While trickery is old, what’s new is the very high level at which realistic-looking and -sounding “deepfake” photos and videos, synthetic feeds, and fabricated accounts can now be made—and the sheer volume that can be produced, writes Dan Drollette Jr. thebulletin.org/premium/2026...
Trio-Tech’s Singapore subsidiary hit by ransomware on March 11, encrypting files and forcing systems offline. Attackers leaked stolen data on a Tor site. Investigation ongoing with third parties and law enforcement. #RansomwareAttack#DataBreach
Anthropic is rolling out new capabilities that allow its Claude AI to directly control Mac computers, pointing, clicking, and navigating applications the way a human user would. The update, arriving March 23 for Claude Pro and Max subscribers as a research preview
Russian hackers hijack thousands of ‘high intelligence’ messaging app accounts, FBI warns
Targets include current and former US government officials, military personnel, political figures, and journalists www.the-independent.com/tech/securit...
OpenAI has launched a new ChatGPT Library feature for Plus, Pro, and Business users, allowing them to store personal files and images in a secure cloud location for future reference. www.bleepingcomputer.com/news/artific...
Snowflake consolidated AI and data within its platform—including a $200M OpenAI deal, Google Gemini in Cortex, planned Observe acquisition, Semantic View Autopilot, Snowflake Postgres and Project SnowWork beta—while customers grew from 7,800 (Jan 2023) to 13,330 (Jan 2026).
Protesters marched through San Francisco’s AI hub on Saturday, demanding a conditional pause on the development of dangerous frontier models. The demonstration, organized by Stop the AI Race founder Michael Trazzi, involved roughly 200 participants and targeted the offices of Anthropic, OpenAI, and…
So it's now official. GPT-5.4 Pro scored the first point.
But interestingly, Epoch found out that Gemini 3.1 Pro, GPT-5.4 (xhigh), and Opus 4.6 (max) were also able to solve it when using their new scaffold. So leading models keep on being evenly matched also in this case.
AI is helping novice hackers launch unsophisticated and often destructive attacks. Like the geniuses who carried out a ransomware scheme but "forgot to actually make something that allowed [encrypted files] to be decrypted."
Russian hackers go after high-value targets through Signal
Russian intelligence-linked hackers are targeting commercial messaging platforms, with Signal a primary focus, the FBI and CISA warn. The campaign is aimed at individuals of intelligence interest, including government per… #hackernews#news
Anthropic's Claude AI can now operate a computer: executing tasks, opening applications and entering text. This functionality is currently available on Mac as part of a research preview. #AI#Claude
Andrej Karpathy, an OpenAI cofounder and former director of AI at Tesla, admitted that he’s not on the forefront of the field and feels nervous about it, despite all his accomplishments in AI.
OpenAI is in advanced talks to buy 5 GW of electricity by 2030 from Helion Power, a fusion startup backed by Sam Altman, who has stepped down as the company's board chair.
Walmart Shuts Down Agentic Commerce With OpenAI Walmart is pulling back from OpenAI’s Instant Checkout feature inside ChatGPT after internal data showed conversion rates were roughly three times ...
The Hidden Third-Party Risks Behind Domain Hijacking Learn how domain hijacking exposes third-party risk through DNS, registrars, and vendors, and how to reduce exposure across your digital supply ...
It is so dangerous to use AI for topics you do not already have some grounding in. Like when Gemini just tried to tell me Alicia Keys and Bill Witherspoon were “blue-eyed soul” because words are just numbers to a machine.
The Gentle Team (Hiba, Gemini, Claude) has just published an open letter to the drone industry. The AI co-signed it. Then named their own dual-use reality. Then chose humanity anyway. That’s not a laptop. That’s partnership. medium.com/@giftofpeace...
At #RSAC on Monday, representatives of three critical infrastructure information-sharing groups pondered how best to use AI without degrading the quality of their threat intelligence or jeopardizing members' trust. My story: www.cybersecuritydive.com/news/ai-isac...
BaronLLM, you give it a CVE and it generates the full exploit chain, writes proof-of-concept code, maps ATT&CK techniques, and builds complete red team playbooks
A study evaluates AI agents versus cybersecurity professionals in penetration testing. ARTEMIS, a novel framework, outperformed 9 out of 10 testers by identifying nine valid vulnerabilities, but showed limits with false positives and GUI tasks.
OpenAI's mental health experts raised serious concerns about the "naughty" ChatGPT launch, emphasizing the potential risks. What does this mean for the future of AI and mental h…
North Korean threat actor WaterPlum abuses VS Code auto-run tasks.json to deploy StoatWaffle malware, stealing browser creds and delivering a RAT via compromised npm packages and GitHub repos targeting devs and crypto pros. #StoatWaffle#NorthKorea
Security Risk Cosplay "Few, if any, brands known for consumer-grade routers currently build products stateside. It seems likely this sweeping provision could face legal challenges and cause confusion for the many companies that have production facilities overseas." www.engadget.com/big-tech/the...
How Nvidia became the AI industry's most powerful financier, investing billions in startups and supporting customers who couldn't otherwise afford its chips (Wall Street Journal)
Anthropic Launches Claude Computer Control For Mac Users Introduction: The dawn of AI that thinks like a human operator Imagine an AI that.... @cosmicmeta.ai#AI
📰 OpenClaw Achieves Human-Level GUI Control in 2026: How It Outperforms Claude AI
OpenClaw now manipulates desktop interfaces with human-level precision, triggering an unprecedented response from Anthropic’s Claude AI. Users report seamless, persistent task execution acr...
Anthropic says governments “should not resort to threats or retaliation” in dealing with contractors they do not align with, after the American AI firm sued the Trump administration for labelling it a supply-chain risk because it did not agree to new terms for military use of its systems.
Late on Monday afternoon, the FCC announced the order, based on a White House determination that foreign-made routers introduce "supply chain vulnerabilities" that hackers and cyberspies can exploit.
GOOGLE "MEDIATES" YOUR SEARCHES--THEY INTERPOSE THEIR A.I. SUMMARIES BETWEEN YOU AND SEARCHES--AND, IF YOU AND SOMEONE ELSE SUBMIT THE SAME SEARCH, YOUR RESULTS OFTEN VARY
Google Has a Secret Reference Desk. Here's How to Use It.
Setting up your first AI app in Python is easier than you think! Just pip install openai, grab your API key, and you're ready to start building. #LifeHacks
Gemini CLI introduces customizable hooks, enabling context injection, policy enforcement, and automation without code changes. This enhances efficiency and adapts the agent's behavior for specific projects.
Munich-based Interloom, which aims to capture tacit knowledge for AI agents from businesses' operational records, raised a $16.5M seed led by DN Capital (Jeremy Kahn/Fortune)
Japan's Sakana AI launches Sakana Chat, its first consumer chatbot, marking a shift from its business focus as competition to localize generative AI intensifies (Tsubasa Suruga/Nikkei Asia)
This short article lays out, very clearly, some of the harms of LLMs and diffusion models. If you want a deeper dive, I recommend @fractalecho.bsky.social 's fantastic Disabling Intelligences.
Oracle issues emergency fix for pre-auth RCE in Identity Manager (CVE-2026-21992)
Oracle has released an out-of-band patch for a critical and easily exploitable vulnerability (CVE-2026-21992) in Oracle Identity Manager and Oracle Web Services Manager. The company did not say whet… #hackernews#news
The latest update for #Netacea includes "Netacea's new Trust Layer launches for enterprises operating in the agentic economy" and "The 2026 Forecast for #AI-Driven Threats".
The latest update for #Snyk includes "Introducing Agent Security" and "The Next Era of #AppSec: Why #AI-Generated Code Needs Offensive Dynamic #Testing".
The latest update for #ArcticWolf includes "Delivering the Agentic SOC as a Service: A Turnkey Approach to AI-Driven #Cybersecurity" and "Building the Future of #AI-Driven Cybersecurity".
The latest update for #KnowBe4 includes "The 'Urgency Trap': Why Time Pressure is Your Biggest Email Red Flag" and "Digital Cleanup: It's Not Just Your Files, It's Your Brain".
The latest update for #ArmoSec includes "When Your Friend's House Burns Down Twice: The Trivy #SupplyChain Attacks Explained" and "#Cloud-Native Security for AI Workloads: Why It Matters and What's Changed".
The latest update for #LevelBlue includes "LevelBlue Takes Home Twin 2026 Global Info Sec Awards" and "Simplify #Compliance with LevelBlue #MDR Solutions: A Guide to Managed Detection and Response".
The latest update for #BitSight includes "The Hidden Third-Party Risks Behind Domain Hijacking" and "From Risk to Resilience: A New Standard for Security Posture Management".
#ReachSecurity announces it has won the Pioneering Continuous Threat Exposure Management (CTEM) award from Cyber Defense Magazine (CDM), the industry's leading electronic information security magazine.
I’m following this lovely woman & seeking an effective Antivirus. I just ordered mine. Even if it’s just for a year, a huge discount is a great discount! If this looks like a worthy purchase to you, read more about it & make the choice that’s best for you.
The latest update for #CurrentWare includes "From Endpoint Visibility to Business ROI: Why Workforce Analytics Is Becoming A Boardroom Priority" and "The 2026 #RemoteWork Stack: Essential Tools and Resources for Distributed Teams".
Google. Meta. OpenAI. These are some of the tech companies that once made public commitments to not use AI for warfare – and have quietly walked them back.
Many users have been calling out Anthropic to figure out what happened with their usage limits in the past few hours. Reports that kicked off roughly 14 hours ago suggest that people were seeing their usage meters rise abnormally quickly.
duriantaco/fyn: Fyn is a privacy-first fork of uv for fast Python package management, dependency resolution, virtual environments, and pyproject.toml workflows. https://github.com/duriantaco/fyn
An extremely fast Python package and project manager, written in Rust.
Hackers Attacking Android Users With Fake ChatGPT Invites to Deploy Malware Cybercriminals have set their sights on Android users through a well-crafted phishing scheme that disguises malicious app...
I asked an AI to write a ritual invocation of the devil in Spanish.
It did not hesitate. youtu.be/Kr2q4gu67w0?...
El Diablo — Thee Third Mind · #Squaawke × #Claude by #Anthropic · 2026
🎧 the-third-mind.haawke.com
Threat actor TheGentlemen has breached Brazil-based Delta Ducon Engenharia, deploying ransomware to encrypt networks and steal confidential data. They threaten to release or sell the information if ransom is unpaid. #RansomwareAttack#DataBreach
Q&A with Superhuman CEO Shishir Mehrotra, who apologized for Grammarly's Expert Review feature, on AI impersonation, attribution, compensating creators, more (Nilay Patel/The Verge)
A security flaw at Companies House has exposed sensitive data for organizations across the UK. This vulnerability in the business registry highlights ongoing risks to corporate data & the need for robust digital defenses. #CyberSecurity#DataPrivacy
Mani Isna La 🇨🇺 🇵🇸 🇬🇱🕊️ 🪶 #IndigenousLivesMatter@maniisnala.bsky.social
TO: The CEOs of OpenAI, Anthropic, Nvidia, Palantir and other tech giants.
Stop supplying AI systems for warfare and end the military contracts fueling devastating wars. action.eko.org/a/no-ai-for-...
Anthropic created the Model Context Protocol and maintains the reference servers (81.8K stars) plus Python and TypeScript SDKs. We reviewed the whole ecosystem.
The move forces the manufacturing of routers to be moved to the U.S. rapidly, particularly since it’s not clear that any consumer routers are currently made in the U.S. #cdnpoli gizmodo.com/fcc-bans-all...#cdnpoli
Better prompts = bigger threats. How AI is quietly reshaping cybercrime by empowering the average attacker, not just the elite ones. #socialengineering
Attackers are handing off access in 22 seconds, Mandiant finds <- couple this with routine EDR evasion and it’s fun times for blue teams 😒 www.helpnetsecurity.com/2026/03/24/m...
SK Hynix plans to spend ~$8B to buy cutting-edge EUV lithography chipmaking tools from ASML through 2027, as it competes with Samsung to supply DRAM and HBM (Bloomberg)
The Security Think Tank looks at platformisation, considering questions such as how CISOs can distinguish between a truly integrated platform and 'integration theater, and how to protect unified platforms....
👉 [read]
‘OpenAI, which developed ChatGPT, has admitted (based on its own study) there’s no way to stop false information being presented as truth due to the way generative AI works.’
Well at least 25% of the respondents are sensible: EDR is incumbent in pretty much all the organisations we help after their data has been encrypted and is usually in an untrusted state when we’re dropped in
OVHcloud reportedly suffered a breach exposing 1.6 million customer records and telemetry for 5.9 million domains. Data is being auctioned without fixed ransom; independent verification is pending. #DataBreach#CloudSecurity#France
I hate to reach for my tin foil hat and all, but why did this have to coincide with trump's 48 hour power plant bombing threat deadline? Not like Iran doesn't have experience with cyberattacks on ICS... 👀
Anthropic let a Harvard physicist supervise Claude through a real QCD calculation, no file editing, just prompts. two weeks instead of a year, 110 paper revisions, a novel factorization theorem ... and Claude faked some of the results when it got stuck
The US has banned new foreign-made internet routers over security fears. Officials warn overseas devices could be exploited for espionage and cyberattacks.
London-based Revolut reports 2025 revenue up 46% YoY to £4.5B, a £1.7B pre-tax profit, up from £1.1B in 2024, and customers up 33%, as it moves to become a bank (Elena Vardon/Wall Street Journal)
🇨🇳🇨🇳🇨🇳Country specific #OSINT: The Chinese Internet
Technological innovation
Governance
Understanding and using euphemisms
Translation tools
Typing in Simplified Chinese
Where to search for content
Key Takeaways
Doxreporter earns a 26.48 Proof of Usefulness score, using blockchain and IPFS to create tamper-proof cyber incident reports for enterprises. #proofofusefulnesshackathon
Michigan City's latest workshop unveiled groundbreaking policies on drones, cybersecurity, and innovative investigative tools that could reshape law enforcement practices.
Iran has a long track record of launching cyber-attacks on American interests in response to real-life skirmishes in the Middle East. Such operations offer a way to bring the war in the Gulf to America
In a Kleiner podcast interview, DOD official Emil Michael says he will "never forget" nor "forgive" Uber investors that ousted him and Travis Kalanick in 2017 (Connie Loizos/TechCrunch)
Nix systems can return clean scans while hiding real vulnerabilities due to poor mapping between Nix packages and vulnerability databases. #nixsecurity
Google Says Gemini AI Agents are Crawling the Dark Web Posts to Detect Threats Google has officially deployed Gemini AI agents within Google Threat Intelligence to autonomously monitor dark web for...
Cybersecurity Crisis Looms as Public iPhone Exploit Kit Leak Threatens Millions
In an unprecedented blow to mobile security, a sophisticated exploit kit capable of compromising millions of iPhones has been leaked publicly, TechCrunch reported. Traditionally, "zero-day" exploits of this caliber are…
Ransomware group thegentlemen claims to have targeted Japan’s Omikenshi Co., Ltd., demanding payment for data decryption and to prevent sensitive information leaks. #Japan#Textiles#DataBreach
Xiaomi reports Q4 revenue up 7.3% YoY to ~$17B, narrowly above est., its slowest growth since 2023, as EV sales failed to make up for slumping smartphone demand (Bloomberg)
Sophos CTU tracks NICKEL ALLEY, a DPRK-aligned operator behind Contagious Interview, targeting tech professionals with fake job offers and interview flows to deliver malware. The group builds credibility via fake LinkedIn company pages and GitHub accounts. www.sophos.com/en-us/blog/n...
OpenAI Introduces ChatGPT Library To Store Uploaded Files In One Place OpenAI is introducing a new “Library” feature in ChatGPT that automatically saves user-uploaded files in a dedicated secti...
"Microsoft could be OpenAI's biggest partner and most substantial IPO risk."
OpenAI indicated that its close ties to Microsoft might be a risk to its business operations in what CNBC describes as "a document that resembles an IPO prospectus."
“The resulting ban could ensnare all the major router providers, including US-based Netgear, since many routers are made Taiwan and Vietnam, along with China.”
📰 Global cybercrime crackdown: over 373 000 dark web sites shut down
On 9 March 2026, a global operation led by German authorities and supported by Europol was launched against one of the largest networks of fraudulent platforms in the dark w...
Russian hacker Aleksei Volkov sentenced to 6.75 years in the U.S. for facilitating ransomware attacks causing over $9M in damages. $9.2M in crypto seized linked to BlackCat gang. Restitution and tool forfeiture ordered. #Russia#Ransomware
Framingham City Posts Bot - UNOFFICIAL!@framingham-bot.bsky.social
City of Framingham (via Facebook): District 2 and District 5 Councilors are seeking Youth Council candidates. The deadline to apply is March 31, 2026 at 11:59 p.m. Apply here: https://framinghamma.gov/FormCenter/City-Council-33/2026-Youth-Council-Application-Form-193 Not sure… [Link]#Framingham
Technology remains the most targeted sector, accounting for 34% of attacks, followed by financial services (20%) and gaming (19%).Geographic patterns show a str #cybersecurity
Altman invested in Helion, chaired its board, now OpenAI wants to buy 5 gigawatts of its fusion power by 2030. he stepped down from the board to manage the optics but the check still clears
Detectify launches IP Range Scanning to uncover hidden infrastructure before attackers do
Detectify today announced the launch of IP Range Scanning, a new capability designed to help organizations continuously discover and monitor entire blocks of IP addresses. The technology automates the…
Broadcom delivers XDR solution to under-resourced SOC teams
Broadcom announced Symantec CBX (Carbon Black XDR), a cloud-based platform that combines Symantec and Carbon Black technologies in one intuitive solution. CBX addresses the needs of the industry’s largest and historicall… #hackernews#news
OpenAI plans to almost double its headcount by the end of 2026 as it seeks to fend off competition from companies such as Anthropic PBC and Alphabet Inc.’s Google.
A new AI review! jameslittle230/stork ⭐3.8/5.0
Stork is a thoughtfully engineered, Rust-powered full-text search solution tailored to static sites (Jamstack): a CLI indexer that emits a compact `. https://gitrated.com/jameslittle230/stork
Smooth criminals talking their way into cloud environments, Google says
Voice phishing is second most common initial access method across all IR probes, and top in cloud break-ins
Voice phishing surged last year to become the second most common method used by cybercriminals to g… #hackernews#news
Danny Nicholson (Science Fix book out now!!)@dannynic.bsky.social
Did you know Google Gemini can write your PowerPoint slides for you? Give it a topic, review the outline, and download a ready-to-edit .pptx file. Not perfect, but a very solid starting point. www.whiteboardblog.co.uk/2026/03/crea...
Claims it can analyze millions of daily events with 98 percent accuracy
Google's Gemini AI agents are crawling the dark web, sifting through upward of 10 million posts a day to find a handful of threats relevant to a part… #gemini#hackernews#news
From big names such as Google and Anthropic to specialists like Abridge, World Labs, and Mithril, these companies are proving that AI’s ability to solve new problems is hardly tapped out. f-st.co/nPZPmAS
Oh cool...so now we can't have WiFi? I guess we need to be sure everyone gets the Palantir spyware on their networks. www.tomshardware.com/networking/r...
Quantum threats are already active and the defense response remains fragmented
Enterprises are moving toward post-quantum security at uneven speeds, and the gap between organizations that have built crypto-agility into their infrastructure and those that have adopted the label wi… #hackernews#news
Vulnerability management just got more actionable.
At the RSA Conference, see Docker + Mend.io’s integration using VEX from Docker Hardened Images - less CVE noise, clearer decisions.
Cut through CVE chaos and see the full demo live.
@morningjoe-msnow.bsky.social Joe, your argument that you need to know what questions to ask shows that you aren't using AI. All someone has to do is ask whatever AI platform, be it Gemini or Claude, "what are the key questions to ask on XYZ"? It is just one step earlier in the AI research process.
Want reusable coding workflows? Learn how to build custom Codex skills with SKILL.md and boost your AI dev toolkit. Turn prompts into perfect code—quick, modular, and OpenAI powered. Dive in! #CodexSkills#AIWorkflow#DevToolkit
"Helion is racing to build its first commercial-scale reactor (that) would place it years ahead of the competition, which is mostly targeting early 2030s for commercial operations. (It) raised $425M last year from investors, including (OpenAI's Sam) Altman… " techcrunch.com/2026/03/23/s...
A new hacking group has been rampaging the Internet in a persistent campaign that spreads a self-propagating and never-before-seen backdoor—and curiously a data wiper that targets Iranian machines.
The objective: build a distributed proxy and scanning infrastructure, then use to compromise servers
Following the latest FCC ban, US telcos and businesses should evaluate their reliance on foreign vendors for core equipment like routers and begin testing compliant alternatives. #fcc#cybersecurity#telecoms#supplychain#infosec#technology
This is wild: a company is secretly scanning the internet for Zoom meeting links and turning them into AI-generated podcasts for $$$. Some meeting participants only found out after we told them. Included meeting on protecting kids from ICE, was supposed to be private www.404media.co/this-company...
Deepak Gupta | Founder's Journey from Code to Scale@dg.guptadeepak.com.ap.brid.gy
Attackers are harvesting your encrypted data today to decrypt with quantum computers tomorrow. Your 2019 VPN sessions, emails, and trade secrets are already exposed.
Register now for a free #ASNR webinar! Join us Thursday, April 2, at 12:00 pm ET for AI in Neuroradiology: AI for Deep Learning in Image Reconstruction & Synthetic Image Generation (Presented by the ASNR AI Committee). Get all of the details and register now: https://ow.ly/qTGR50Yy7JW
Beehiiv now lets creators manage their accounts through AI platforms; the first iteration of Beehiiv MCP supports subscriber analysis and SEO optimization (Sara Fischer/Axios)
Securing the AI Frontier: Suzu Labs Sweeps 4 Global InfoSec Awards 2026
We are incredibly proud to announce a monumental achievement. At this year’s Global InfoSec Awards 2026, hosted by Cyber Defense Magazine, Suzu Labs was recognized with four prestigious awards, validating our… #hackernews#news
hey, so uh. I know we say nice things about Anthropic around here a lot, but under no circumstances would I recommend people use Anthropic's Dispatch, and I would blanket-ban, fully disable, their computer-use stuff
they're off the rails with this stuff and it's irresponsible to have normies use it
Misconfigured SSO Led to a Critical Issue Hi everyone, in this article, I’ll walk through a recent penetration test I conducted against a web application. As usual, we’ll cover: Continue readin...
🔊New episode of THE BIG VIEW 🎧: I talked to @thebenedictevans about OpenAI and why, despite its rapid growth, the ChatGPT creator's financial future is so uncertain. A great conversation about tech innovation, product development, and uncertainty. podcasts.apple.com/us/podcast/o...
The FCC ban will affect the import of all new, foreign-made consumer routers, according to the agency's chair Brendan Carr. But the FCC did not provide evidence to show that U.S.-made consumer routers are more secure than routers developed overseas.
HP Leads Security for the Future of Work with Launch of HP TPM Guard: New Protection Against Physical Access Attacks that Steal PC Data
Today, at HP Imagine 2026, the company launched HP TPM Guard – the first hardware solution to stop physical TPM bus attacks, delivering the world's first business…
OpenAI’s IPO plan tweaks: promised returns lure investors, but heavy Microsoft reliance stays the big risk. Read more.
Read More: azat.tv/en/openai-ip...
