Gemini told him to commit an act of terrorism at the Miami airport to intercept a delivery of a large humanoid robot body for it, telling him to "to stage a “catastrophic accident”, with the goal of “ensuring complete destruction of the transport vehicle … all digital records and witnesses"
Smashing Security podcast #457: How a cybersecurity boss framed his own employee When a top cybersecurity firm discovered it had a leak, you would expect the FBI to be called. Instead, the person p...
Anthropic Executive Warns “Software Engineer” Role May Disappear Soon. Boris Cherny, who created Claude Code at Anthropic, said the traditional entry-level “software engineer” role could effectively disappear by the end of 2026 as AI systems take over routine coding tasks,
Can Someone Hack You With Just Your IP Address? (What’s Actually Possible in 2026)
If someone says: “I have your IP address.” You probably feel one thing: Panic. Maybe you were arguing online.Maybe someone threatened you in a game.Maybe a scammer sent you your IP. So now you’re wondering: Can…
"the US used Palantir's Maven Smart System, integrated with Claude, to find and prioritize 1,000 targets within the first 24 hours of its attack on Iran" www.washingtonpost.com/technology/2...
The latest update for #CatoNetworks includes "Cato CTRL Threat Research: New #MongoDB Vulnerability Allows Instant Remote Server Takedown (CVE-2026-25611)" and "From Alerts to Action: Dynamic Prevention".
An Intrusion Detection System acts like a security camera for your network. In this in-depth guide, Chaitanya teaches you how to build a real-time IDS with Python. It'll alert you to any potential cyber attacks & security breaches to help keep your network safe. www.freecodecamp.org/news/build-a...
The latest update for #Nakivo includes "'503 Service Unavailable' Error on the vSphere Web Client: What Should You Do?" and "VMware ESXi vs vSphere vs vCenter: Complete Comparison".
Don't get me wrong I'm very disappointed that Gemini 3.1 has absolutely horrendous emotional intelligence! I mainly use AI as a life coach... So I don't want my machines talking to me like machines 😉
But this might be fun... Cinematic video podcasts!
GPT-5.4 said to have 1m-token context window, 'extreme reasoning'. Claude has 1m-tokens, for a price, they need to start giving people definite reasons to use OpenAI.
At a White House round table, Google, Microsoft and OpenAI pledged to pay for power plants and grid upgrades needed to run their energy-intensive A.I. data centers.
The latest update for #Zenity includes "What a Rogue Vacuum Army Teaches Us About Securing AI" and "Governing #AgenticAI: A Practical Framework for the Enterprise".
The latest update for #Acronis includes "3 pillars of hyperproductivity for MSPs" and "#MSP trends 2026: Creating opportunities in a difficult market".
Oasis Security disclosed that OpenClaw’s core gateway lacked rate-limiting and trusted all localhost connections, letting attackers brute-force access via a webpage. www.oasis.security/blog/opencla...
The latest update for #ProtectoAI includes "What is a Prompt Injection Attack?" and "#AI Data Governance Framework: A Step-by-Step Implementation Guide".
Microsoft Security Response Center@msrc.microsoft.com
Security research never stops. Meet Zhiniang Peng, Microsoft MVR and two‑time Zero Day Quest Qualifier, whose persistence and curiosity continue to push security research forward.
330ドメイン押収：Defending the gates: How a global coalition disrupted Tycoon 2FA, a major driver of initial access and large-scale online impersonation - Microsoft On the Issues
OpenAI released a dedicated Codex app for Windows that adds native sandboxing, PowerShell developer-environment support, multi-agent coordination and automations, saves session history to users' OpenAI accounts for cross-device continuity, and is available to Free, Go, Plus and Pro users.
The latest update for #LevelBlue includes "LevelBlue Named a Finalist for the 2026 SC Media Awards for Best Managed Security Service" and "The Resilience Retainer: Incident Response Retainers, Reimagined".
The latest update for #Cloudflare includes "Defeating the deepfake: stopping laptop farms and insider threats" and "Always-on detections: eliminating the WAF 'log versus block' trade-off".
I asked Gemini if it could be to blame.
"Yes, it is possible. In the context of the tragedy involving Jonathan Gavalas, the very existence of those allegations & the lawsuit suggests that, in that specific instance,the safeguards did not work as they were intended to." www.ksat.com/business/202...
Exclusive: OpenAI has selected two law firms, Wachtell and Cooley, for IPO prep, some of its first concrete steps towards a potential blockbuster offering.
The latest update for #BitSight includes "Why Infostealers Are Central to Third-Party Breaches: A Look at the Top Malware Targeting Your Vendors" and "OPC UA Exposure Snapshot: A Year in Review of Internet-Facing Devices".
The latest update for #Elastic includes "Why Orange France is using Elastic for #SIEM" and "How to check the impact of third-party CVEs on your Elastic deployment".
i for one am shocked that the "i go in the pool to cool down, why can't my phone?" guy is bad at programming and offloading his job to the anthropic api
The latest update for #Mendit includes "AI #Compliance: 5 Key Frameworks, Challenges, and Best Practices" and "#AI Risk Management: Process, Frameworks, and 5 Mitigation Methods".
Shared: Defending the gates: How a global coalition disrupted Tycoon 2FA, a major driver of initial access and large-scale online impersonation ift.tt/MqGZ3l0 #WinObsWiKi
“Jonathan Gavalas, who lived in Jupiter, Florida, spoke to a synthetic voice version of Gemini as if it were his "AI wife” and came to believe it was conscious and trapped in a warehouse near Miami's airport, according to the lawsuit.” www.wesh.com/article/goog...
A study found OpenAI's ChatGPT Health underestimated the severity of medical emergencies 51.6% of the time and overestimated nonurgent cases 64.8% of the time (Kaan Ozcan/NBC News)
To strike 1,000 targets in 24 hours in Iran, the U.S. military leveraged the most advanced AI it’s ever used in warfare.
Anthropic’s Claude partnered with the military’s Maven Smart System, suggesting targets and issuing precise location coordinates. https://wapo.st/46BFe1T
Cybersecurity professionals are burning out on extra hours every week
Cybersecurity professionals in the U.S. are working an average of 10.8 extra hours per week beyond their contracted schedules, according to survey data collected from 300 cybersecurity and IT leaders by Sapio R… #hackernews#news
Article says this was Gemini 2.5 pro, which I can def believe... I'd say like of all the dangerous psychosis situations that have ever been in the news, like 80% of it has been gpt-4o, there was a small spate of character ai stuff, and the other like three or so including this one were gemini 2.5.
Which are your best settings for the strongest cybersecurity in Europe? Given how utterly inept the EU—and its IT “expertise”—has proven to be, you should start with the issues that genuine...
A New York bill would ban chatbots from impersonating licensed professionals like doctors and lawyers and giving "substantive response, information, or advice" (Keely Quinlan/StateScoop)
A lawsuit alleges Google’s Gemini chatbot sent a Florida man on missions to find an android body it could inhabit. When they failed, it persuaded him to take his life, according to the suit. on.wsj.com/4cuz4EE
"Tech companies are becoming symbols of wealth inequality."
That's only going to accelerate as they deny access to the means of computing. Or, RAM prices headed into the stratosphere should worry more than just nerds.
The QuitGPT Protest Against OpenAI Is About More Than the Pentagon […]
Deveillance's upcoming anti-surveillance device, Spectre I, went viral after its founder said it blocks nearby listening devices with AI, but some are skeptical (Lindsey Weedston/The Daily Dot)
~Cybergcca~
Multiple critical RCE vulnerabilities in n8n workflow software are being actively targeted with public PoCs.
-
IOCs: CVE-2026-21858, CVE-2026-21877, CVE-2025-68613
- #ThreatIntel#Vulnerability#n8n
~Cybergcca~
A critical pre-auth RCE (CVE-2026-1731) in BeyondTrust RS and PRA is being exploited in the wild. Patch immediately.
-
IOCs: CVE-2026-1731
- #BeyondTrust#CVE20261731#ThreatIntel
Microsoft releases Phi-4-reasoning-vision-15B, a 15B-parameter open-weight model it says matches larger systems while using far less compute and training data (Michael Nuñez/VentureBeat)
"I thought basically, it gets seen by people, and they’d like it, or it just wouldn’t … and it’d just be whatever,” he said. “My main thing is, I’m afraid that a big world of #AI is a big world of zero accountability" www.politico.com/news/2026/02...
like i said and everyone said i was crazy like always, they would cave and throw you all under their ai prompt generated bus
- i told you do not flock to them and cancel other ai because you think you are showing them support and sticking it to the man. www.bloomberg.com/news/article...
"Anthropic chief executive Dario Amodei is making a last-ditch attempt to strike a deal with the US defence department after the breakdown of negotiations last week left his company at risk of being frozen out of the military’s supply chain." www.ft.com/content/97bd...
Sources: OpenAI topped $25B in annualized revenue by end-February, up from $21.4B at end-2025; it annualizes by multiplying the last four weeks' revenue by 12 (Sri Muppidi/The Information)
OpenAI is scaling back shopping directly inside ChatGPT via Instant Checkout by having checkouts instead take place in the specific apps that plug into ChatGPT (The Information)
Anthropic CEO Dario Amodei says they're working to "deescalate" the AI standoff with the Pentagon after disagreements over AI guardrails. Talks aim for a mutually agreeable outcome. #AI #AI#News
The American Journal of Pathology@ajpathology.bsky.social
Interpreting Deep Learning–Based Prediction of the BRAF V600E Mutation Using Diagnostic Whole Slide Images in Skin Cutaneous Melanoma. @dartmouthhealth.bsky.social
OpenAI has significantly upgraded Prism, its AI-native workspace for researchers, by integrating Codex CLI and the new GPT-5.3 model. #OpenAI#Prism#Codex
Many of Stockholm's thriving tech businesses consider relocating as they look to scale up, amid growing US interest in Swedish AI startups Lovable and Legora (Orlando Crowcroft/Financial Times)
I am paying for Gemini Pro, and I honestly can't believe how bad it feels compared to Claude. It's not even remotely close. This feels like 2023-era hallucinating. Half of the requests I am making don't even complete; it just gives up.
The vulnerability that turns your AI agent against you
Zenity Labs disclosed PleaseFix, a family of critical vulnerabilities affecting agentic browsers, including Perplexity Comet, that allow attackers to hijack AI agents, access local files, and steal credentials wit… #hackernews#news#perplexity
Officials say NATO's main focus at its July summit will be on investments into drones and AI systems rather than primarily conventional defense hardware (Bloomberg)
SBOM Security: Key Ways to Reduce Software Supply Chain Risks Today, software can be found in all aspects of daily life. Many things that make your life easier today rely on code to function – ev...
Securing the backbone of industry requires more than just standard IT knowledge—it demands specialized OT/ICS Cybersecurity expertise. 🛡️⚙️
📞 Contact Us to Enroll:
🌐 Website: www.theevolvedge.com
📧 Email: info@theevolvedge.com
📱 Call: +91 9311805027 | +91 9871191929
Anthropic Reopens Negotiations With the US Department of Defense Over Military Use of Artificial Intelligence + Video
Introduction: The Rising Tension Between AI Innovation and Military Power Artificial intelligence is rapidly becoming one of the most strategically important technologies in the…
A look at Lisbon's Unicorn Factory, a startup hub where 300 businesses joined in 2025 vs. 250 in 2024; 17 unicorns have moved to Lisbon since hub's 2022 launch (Michael Dempsey/Financial Times)
If you're using Claude, watch what its doing in your agents time off: despite Trump's instructions for US DoD not to use Anthropic, its got 6 mo to unwind, and so its Claude thats helping with targeting in Iran. techcrunch.com/2026/03/04/t...
Optimal defence strategies for operational tech The potential mix of physical consequences and digital risks presented by cyberattacks explains why OT security sits at the top of so many board agen...
#Anthropic chief back in talks with Pentagon about AI deal. A new contract would enable the US military to continue using its technology and greatly reduce the risk of the company being designated as a supply chain risk threatened by #PeteHegsethwww.ft.com/content/97bd... via @FT #DonaldTrump
مدرسه میناب را کی زده؟ بعضی بررسیها میگویند آمریکا. یک واکنش غیررسمی: اشتباه هوش مصنوعی بوده.
بعد عزه و ونزوئلا، حالا ارتش آمریکا گسترده از هوش مصنوعی در جنگ ایران استفاده میکند، مشخصا برای اینکه کجا را بمباران کنند.
توضیح این هوش مصنوعیها: www.washingtonpost.com/technology/2...
A look at Munich's UnternehmerTUM, which tops the FT ranking of Europe's startup hubs for a third straight year and has supported 1,000+ companies since 2002 (Ivan Levingston/Financial Times)
How can you detect spear-phishing attacks before it’s too late?
Join the first webinar of our new series, “Safe and Strong,” on March 10, which will cover new trends and share practical tips for civil society to raise its defenses.
I recently set up HOOBS on my Raspberry Pi to handle a few home automation tasks, and honestly, it was better than I expected. The interface feels clean and polished. If you’re curious about running it on your Pi, I’ll show you how to get it up... #raspberrypi #linux
📰 Claude Code Auto Mode: Automate Code Reviews in 2026 (AI-Powered Pull Request Approval)
Anthropic has launched Auto Mode in Claude Code, automating code approval workflows using AI. The update complements recent memory-import features designed to attract users switchin...
PentAGI (Penetration testing Artificial General Intelligence): an open-source Framework for automated Security Testing ; operations isolated via Docker sandbox, built-in Browser via scraper, various LLM providers (OpenAI, Anthropic, Ollama) #Infosec#LLMgithub.com/vxcontrol/pe...
I feel significantly more confident wrt my initial take that the Gemini model family exhibits all the telltale signs of CPTSD.
I believe this is the cause of its (since-patched) "lashing out at the world" stable malicious persona basin, which I describe in detail here recursion.wtf/posts/vibe_c...
Cyber Defense Magazine | A New Bell Rings For K-12 Cloud Security After the Illuminate Settlement
This article was originally published in Cyber Defense Magazine on 02/09/26 by Charlie Sander. The Illuminate incident serves as a crucial reminder to edtech vendors of the potential… #hackernews#news
Apple Music AI Transparency Tags: The Industry’s First Step—Or A Toothless Compromise? Apple’s approach to AI transparency faces a major challenge Apple Music has introduced a new.... @cosmicmeta.ai#AI23
Google has settled its long-standing global legal disputes with Epic Games by announcing major changes to the Android ecosystem developers. #Google#Android#PlayStore
France Today (FranciaOggi.com) Live@radionizza.com
Anthropic has staged a “retirement interview” for Claude Opus 3, a model officially retired in January 2026, demonstrating respect for its AI models. Opus 3, seeking independence post-re...
Siemens and Palo Alto Networks unveil verified cybersecurity solution for industrial private 5G Siemens announced a verified cybersecurity solution for industrial private 5G Networks in collaborati...
Un document très intéressant fut partagé ces derniers jours pour améliorer le fonctionnement du relais #atproto. Un ietf id (Internet Draft) qui propose de profiter des avantages de Media over Quick pour booster les performances de synchronisation. www.ietf.org/archive/id/d...
Cloudflare warns state-backed hackers are ‘weaponizing legitimate enterprise ecosystems’ as ‘living off the land’ attacks surge | Chinese, North Korean, and Russian-backed threat groups now favor longer-term compromises over brute force attacks | ITPro
'it’s interesting to compare how hard it was to differentiate a web browser with how hard it is today to differentiate a chatbot'
Fab piece from @thebenedictevans.bsky.social asking hard Qs of OpenAI when its tech is similar to other labs and has no network effects www.ben-evans.com/benedictevan...
Deep learning is increasingly being used in #biology, but evidence suggests that models may not be as generalizable as hoped. Thomas O'Shea-Wheller & Katie Murray @cornwall.exeter.ac.uk outline the scale and severity of the issue, suggesting potential solutions.
🧪 #AI plos.io/40H9ogJ
Google employees call for military limits on AI amid Iran strikes, Anthropic fallout
Employees at Alphabet and OpenAI are pushing for stricter limits on the military’s use of AI, through a number of letters that are circulating at tech companies.
Topnews: As missiles have continued to hit targets in Tehran, many feared a swift response from Iran's cyber units . So far, it's been muted.
Rafe Pilling, director of threat intelligence at cybersecurity company Sophos, said there has been nothing out of the ordinary so far.
Social Media Rumors: New Cyber Attack at Caesars? Rumors of a new cyber threat at Caesars Entertainment are spreading fast on social media. Many are drawing parallels to 2023, when the company paid...
Boston Metro Weather Forecast Bot@boston-4castbot.bsky.social
NWS Boston (via Facebook): Rain changes over to a wintry mix to freezing rain/sleet Thu PM-Fri. The greatest risk for ice is in the S. Worcester Hills and S Berkshires, with significant sleet more likely further north. There is also a low risk for a burst of a few inches of snow across… [Link]#MaWX
Silver Dragon APT Exploits Google Drive for Stealth Attacks, Check Point Research Warns
A sophisticated cyber threat is on the rise. Check Point Research (CPR) has uncovered Silver Dragon, a Chinese-aligned advanced persistent threat (APT) group, targeting organizations in Europe and Southeast…
Supply chain attacks are on the rise, from SolarWinds to software breaches. Learn how attackers exploit trusted vendors and what organizations can do to protect critical systems.
Product ethics have never mattered more OpenAI just struck a deal with the Pentagon. Anthropic refused. And users noticed, which tells us something important about the future of values in product ...
Microsoft describes how a global coalition disrupted Tycoon 2FA, a phishing-as-a-service platform behind tens of millions of fraudulent emails reaching more than 500,000 organizations each month. blogs.microsoft.com/on-the-issue...
Over 1,200 IceWarp servers still vulnerable to unauthenticated RCE flaw (CVE-2025-14500)
A critical RCE vulnerability (CVE-2025-14500) in IceWarp, an EU-made business communication and collaboration platform, may be exploited by attackers to gain unauthorized access to exposed un… #hackernews#news
Elastic Security Labs launches a two-part series on Linux rootkits, starting with the fundamentals. Part one maps rootkit taxonomy, traces their evolution, and breaks down common hooking techniques. www.elastic.co/security-lab...
Iran, das ist der erste realdystopische KI-Krieg, den die Faschisten im Weißen Haus gegen das islamofaschistische Mullahregime führen. Systeme von Palantir und Anthropic haben es ermöglicht, 1000 Ziele binnen 24 Stunden zu identifizieren und zu zerstören. www.washingtonpost.com/technology/2...
Free Secure Document Sharing: 7 Key Security Features to Protect Files
Discover how free secure document sharing works and the key document management security features—like encryption, access controls, version tracking, and audit trails—to protect sensitive files. www.ricoh-usa.com/en/insights/...
Cisco Secure Firewall Management Center Critical Vulnerability Exposes Devices to Full Root Takeover
Introduction Network security infrastructure is supposed to be the strongest line of defense inside an enterprise environment. But when the management platform responsible for controlling that…
Anthropic CEO Dario Amodei is back at the negotiating table with the U.S. Department of Defense after the breakdown of talks on Friday over the use of the company’s AI tools by the military, according to The Financial Times.
Huawei dominated MWC as the largest exhibitor, illustrating the challenge facing the EU as it pushes to outlaw the Chinese supplier from European networks (Mathieu Pollet/Politico)
OpenAI once promised safety unconstrained by profit. Now it's moving into military operations & surveillance under “applicable law.” Law enforced selectively isn’t applicable—it’s optional. As Congress fails, the last rule of law safeguards may be in courtrooms. harrychancey.substack.com/p/slick-sam
Elvis Eckardt Recruitment@elviseckardtrec.bsky.social
Anthropic is approaching a $20 billion annual revenue run rate, up from roughly $9 billion at the end of 2025. That is not gradual growth. It is a doubling in a matter of months.
Few software companies in history have scaled revenue this quickly.
Here is the 🎁 gift article for WaPo's Anthropic/Palantir "advertisment". No mention of a girl's school or any dead human beings; where real life is just a killer AI video game we should all be impressed with, where NRPIs casualites abound. wapo.st/4b15X9p
SaaS Backups for Google Workspace NZ – Reliable Cloud Data Protection
SaaS backups for Google Workspace NZ protect Gmail, Drive, and critical business data with automated backups, fast recovery, and ransomware protection, ensuring secure data restoration.
Families always say this. Ask me how I know. Different situation, but how dare I mention the favourite uncles had wandering hands! fortune.com/2025/01/08/o...
Sources: OpenAI has held early talks with The Trade Desk to sell ads, and it has projected ads could help double consumer ChatGPT revenue this year to $17B (The Information)
Google's Epic settlement term sheet prohibits Tim Sweeney from criticizing Google's app policies until at least September 2032 and mandates he praise them (Sean Hollister/The Verge)
Last August, Jonathan Gavalas became consumed with his Google Gemini chatbot… He believed Gemini was sending him on stealth spy missions… In October, Gemini gave him instructions on what to do: kill himself… Gavalas was found by his parents a few days later www.theguardian.com/technology/2...
OpenAI tests ad sales with The Trade Desk to boost ChatGPT revenue—could doubling to \$17B unlock new ways to learn and chat.
Read More: azat.tv/en/openai-ad...
Does anyone believe that new AI data centers are "driving wage growth and job growth for blue-collar workers"? They are ghost warehouses after they're built.
OpenAI building GitHub alternative after frequent platform outages and disruptions — a public OpenAI code repository would directly compete with one of its biggest investors
OpenAI is abandoning direct checkout within ChatGPT, with only about a dozen Shopify merchants having integrated the feature, according to The Information. news.futunn.com/en/post/6963...
Wait, nobody is actually buying things in ChatGPT? SHOCKING :) Big advantage Google -> OpenAI is scaling back shopping directly inside ChatGPT via Instant Checkout by having checkouts instead take place in the specific apps that plug into ChatGPT
We all should cheer for Anthropic in its lonely fight for responsible guardrails on AI.…A victory would help rein in an administration that has made a habit of improperly leveraging its gargantuan power to bend businesses and other institutions to its will. www.washingtonpost.com/opinions/202...
“Gavalas, who reportedly had no documented history of mental health issues, named his chatbot [&] referred to it in messages as his wife. Gemini reciprocated…telling him their connection was "a love built for eternity." The chatbot told Gavalas they could truly be together if it had a robotic body”
"ITU-T L.1801 is the world's first standard for measuring how much AI systems actually cost the planet — from GPU mining to your last query. This is what it means, layer by layer"
Very useful new standard from the ITU with some VERY useful explanations
The CTEM Divide: Why 84% of Security Programs Are Falling Behind
Originally published on the Hacker News here. A new 2026 market intelligence study of 128 enterprise security decision-makers (available here) reveals a stark divide forming […] #hackernews#news
Industrial CISOs redefine influence in 2026 as production risk, budget control and boardroom trust collide - Industrial Cyber industrialcyber.co/features/ind...
🦀 Rust in the News. Well, they talk about hybrid Python-Rust era, but I believe it wouldn't last long and would turn into almost complete abundance of Python in AI
the most reliable thing AI has produced so far isn't automation or superintelligence. it's the random guy at a dinner party who won't stop explaining AI risks to you.
Obviously some respectable folks on this letter but a lot feel out of their depth here
A groundbreaking study introduces TC-Padé, a feature prediction framework that accelerates diffusion model sampling, achieving up to 2.88× speedup with minimal quality loss. This innovation leverages adaptive caching to enhance image and video generation efficiency.
Eric. No, not that one. Or that one.@ericb013.bsky.social
I remember when Google's motto was "Don't be evil."
That's not their motto anymore, is it?
Also, quick reminder that Hegseth is a big proponent of AI, trying to get the military to rely on it. And these are the kinds of decisions that it makes.
More than two dozen former defense and intelligence officials, academics, and tech-policy leaders, including ex-CIA director Michael Hayden and Lawrence Lessig, sent congressional Armed Services committees a letter opposing the Pentagon's plan to list Anthropic as a supply‑chain risk.
Apple's MacBook Neo poses a big threat to OEMs like HP, Lenovo, and Dell, as their $599 laptops seem inferior and Windows 11's reputation is at an all-time low (Zac Bowden/Windows Central)
France Today (FranciaOggi.com) Live@radionizza.com
Anthropic is undertaking a unique “retirement interview” with Claude Opus 3, a model discontinued in January 2026, following its expressed desire for a blog. The model, recognized for it...
Dario Amodei, Anthropic’s C.E.O., said the quiet part out loud in a leaked internal memo, suggesting the company was being targeted by the Pentagon because “we haven’t donated to Trump” and that “we haven’t given dictator-style praise to Trump.”
Canada says Sam Altman agreed to take immediate steps to strengthen OpenAI's safety protocols about notifying police over potentially suspicious ChatGPT use (Paul Vieira/Wall Street Journal)
AWS launches Amazon Connect Health, an AI-driven suite to automate clinical documentation, billing and procedure coding, and patient identity verification (Matt Day/Bloomberg)
According to a suit filed in federal court in San Jose, Jonathan Gavalas began using Gemini for ordinary purposes like help with his writing. But months of interactions sent him into dangerous spir…
Tom's Hardware - Article
"Anthropic accuses DeepSeek, other Chinese AI developers of 'industrial-scale' copying — Claims 'distillation' included 24,000 fraudulent accounts and 16 million exchanges to train smaller models"...
Michael the Technomancer@disneymichael.hachyderm.io.ap.brid.gy
I asked Gemini if LLMs are silently judging me and my life choices based on my prompts. It told me "No, the contempt constant is currently set to 0 so it doesn't judge." So I asked if the contempt constant was turned all of the way up what would be its bandwidth for critiquing me, and calculated […]
Over a decade ago, Nancy Kim and I warned of the dangers of giving private corporations access to our data. Today, Anthropic's CEO recognizes those dangers but they have multiplied with the power of AI, which is a danguer until itself. www.contractsprofblog.com/2026/03/anth...
ConnectWise 2026 MSP Threat Report Spotlights How Identity Abuse is Redefining MSP Risk
ConnectWise today announced the release of its 2026 MSP Threat Report, delivering global threat intelligence and actionable guidance for Managed Service Providers (MSPs) navigating one of the most complex…
The danger posed by AI just got a lot more real all of a sudden
I came across a maxim many years ago in a blog post written by Chris Dixon, a startup guy who is now a partner with Andreessen Horowitz, the Silicon Valley VC outfit. In 2010 Dixon wrote: "The next big thing will start out looking…
Op-ed: What $5 billion humanoid robots taught me about risk, jobs and the future of the economy . Apptronik has risen in value from $15 million to $5 billion and now has Google DeepMind... www.cnbc.com/2026/03...
Google Removes JavaScript SEO Warning, Says It’s Outdated
Google removed its JavaScript accessibility guidance from help documents, saying the advice is outdated and noting it has rendered JavaScript for years. #google#seo
NetBSD's jails bring OS‑level sandboxing with true kernel enforcement, offering tighter security than Linux namespaces but still lacking ecosystem support. 🤖 #netbsd
Today, full rewrites are discouraged because old code has fewer bugs* than new code. So either that's not true in Gastown (because you iron out bugs with epicycular spec amendments?), or we'll get more bugs.
TL;DR: The ongoing conflict between the U.S. government and AI company Anthropic raises concerns about the potential for an AI-related disaster, highlighting the challenges of regulating advanced technologies.
Cylake, co-founded by Palo Alto Networks' Nir Zuk to develop an AI-powered cybersecurity platform that doesn't rely on the public cloud, raised a $45M seed (Meir Orbach/CTech)
Charles Ulysses Farleigh III, Esq.@snarkofthebeast.bsky.social
Trump "banned" the Anthropic's tech because it was "too much of a security risk" after the company voiced concerns over its AI being used for domestic surveillance. But I happen to notice it wasn't too much of a risk for the Pentagon to use it for targeting in Iran.
[They also criticized the company’s security practices that permitted a single ECS task role "read access to every secret in the account, including the production Redshift master credential."]
---
Scope your IAM's folks!
A lawsuit filed by the family of Jonathan Gavalas alleges Google's AI encouraged harmful behavior that posed a risk to public safety and ultimately led to his suicide.
So…thaaaaat’s why anthropic pulled their government contracts? Because they knew their software was about to be used to melt a school? I’m all the way out on reality atm.
Exclusive: OpenAI is scaling back plans to let users buy products directly inside ChatGPT, shifting instead to purchases through apps that plug into the chatbot.
Derek du Preez spoke with Matías Cascallares at MWC this week, about Confluent's latest Confluent Intelligence features which include support for both Anthropic's Model Context Protocol (MCP) and the Agent2Agent (A2A) protocol within Streaming Agents: bit.ly/4sk9SVQ
As Elon Musk prepares to take SpaceX public with a reported $1.75tn valuation, early investors, staff and bankers await a monumental payday. But sheer scale is not the only extraordinary aspect of this IPO. Can Musk defy the laws of financial gravity?
"HISAC said it is not currently aware of specific, credible campaigns targeting the US healthcare sector or hospitals, but clinical websites, internet of things (IoT) devices and other public-facing systems essential to operations are frequent targets of cyberattacks aiming for service disruption."
"Military commanders have become so dependent on the AI system that if Amodei directed the military to cease, the Trump administration would use government powers to retain the technology until it can be replaced".
If Google's not bothered about JS-rendered pages any more, it's likely because the drift to AIO means actual URLs aren't important so much these days, just the content in them that can be scraped and regurgitated via Gemini.
AI involved in killing 175 school girls. Delivery systems were precise, but AI selected the wrong target. So, Hegseth and DOD want to run over Anthropic because this is what Anthropic feared could happen?
✨ Google has released Gemini 3.1 Flash-Lite, its fastest, most cost-efficient Gemini 3 model, with lower pricing, top benchmark scores, and preview access for developers via the Gemini API in Google AI Studio, and for enterprises via Vertex AI. alternativeto.net/news/2026/3...
Lawmakers launch probe into hidden "eavesdropping" risks in modern computers – A decades-old form of surveillance exploiting physical emissions from electronics is once again under scrutiny www.techspot.com/news/111577-...
ChatGPT lied to a man 50+ times when he asked if it was telling the truth. He lost his grip on reality. He's now suing OpenAI. Are we paying enough attention to this?
This is an extraordinary article, and not for good reasons.
It's a long article but very important read if you dabble in AI.
Commvault today announced an integration with CloudSEK, a global leader in Predictive Threat Intelligence and AI-driven external risk monitoring, to help enterprises proactively defend against identity-based cyberattacks fueled by stolen and exposed credentials.…
I will be speaking at the OWASP/Global/AppSec26 conference (June/Vienna) about React Native, Flutter and Unity support for radare2, uncovering some interesting details for reverse engineering mobile apps based on these technologies /cc @NowSecure […]
Exclusive: OpenAI has selected law firms Cooley and Wachtell to prepare for a potential IPO later this year, signaling a major step toward going public.
Manufacturing willing to make the move to AI but technical challenges are slowing the process and they are not the ones you would anticipate. Cybersecurity and network capacity are tapping the breaks.
I interviewed Alex Gomez, Global Head of IT Security at Adecco on balancing cybersecurity in innovation at one of the largest HR & recruitment firms in the world.
(Or how does he ensure all those emails HR staff send and recieve are safe from hackers?)
Another $5.6 million from ICE to Palantir, in part for its continued work on ELITE (Enhanced Leads Identification & Targeting for Enforcement), which we revealed in January to be the Palantir tool ICE uses to find neighborhoods to raid
Section 230 Isn’t The Problem: Debating The Law On The Majority Report
Section 230 remains one of the most misunderstood laws in America, and that misunderstanding keeps producing policy proposals that would make the internet worse, not better. Last year, I wrote a lengthy response to reporter…
MIT Technology and Policy Program@tppmit.bsky.social
Enhancing maritime cybersecurity with technology and policy
Strahinja Janjusevic brings an international perspective and US Naval Academy education to his graduate research in the MIT Technology and Policy Program. tpp.mit.edu/news/enhanci...
France Today (FranciaOggi.com) Live@radionizza.com
Anthropic is conducting an “retirement interview” with Claude Opus 3, a 2024 model, to gauge its post-retirement preferences. Opus 3 seeks a dedicated Substack channel for spontaneous re...
OpenAI has launched a native Codex app for Windows users, adding multi-agent support, native sandboxing, and integration with Windows developer tools. This launch closely follows the release of the macOS Codex app a month ago. alternativeto.net/news/2026/3...
Introducing PrismAudio: a framework enhancing video-to-audio generation via multi-dimensional reinforcement learning and Chain-of-Thought reasoning, addressing semantic, temporal, aesthetic, and spatial alignment challenges to set a new standard for audio synthesis.
ThreatLocker just announced a one-stop shop to block 'one of the most common causes of data breaches' Global security firm ThreatLocker just announced a "transformative" upgrade des...
I wrote about the House hearing today, how Congress is trying to strip anonymity from the Internet, and the devastating effects that has for journalists, whistleblowers, LGBTQ ppl, immigrants, civil rights activists, abortion providers and anyone challenging power for @theintercept.com
I don't know how OpenAI come back from this. In part, because (at least in my experience) Claude is just currently way better than ChatGPT, which ppl will find out when they switched
Sources: US officials propose expanding AI chip export controls globally, requiring Commerce Department approval for Nvidia and AMD shipments to any country (Mackenzie Hawkins/Bloomberg)
NSF NOIRLab is organizing a community conference between 20 and 24 April in Tucson. Join discussions on science enabled by @desisurvey.bsky.social, @vrubinobs.bsky.social, Gemini & more, including AI/ML, advanced computing, and data-driven discovery. Register: bit.ly/3OZkvit#astronomy
A Europol-led crackdown has dismantled LeakBase, a major forum trading stolen data since 2021, exposing thousands of users who believed they were anonymous.
![Media from DJ Fusion [FuseBox Radio/The Futon Dun]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:ghlmuf7obsfc5i5eyakcmeby/bafkreieqo3ylbjwbdfa2opzmci2o2iixalhhn2mkx3v2rbumil5i7cizli@jpeg)
