“Defense Secretary Pete Hegseth met with Anthropic CEO Dario Amodei and threatened to invoke the Defense Production Act (DPA) if Anthropic doesn't agree to the Pentagon’s terms by Friday.”
TL;DR: A judge ruled that xAI cannot claim that OpenAI stole trade secrets merely because it hired former employees, deeming their interpretation of text as insufficient evidence.
Smashing Security podcast #456: How to lose friends and DDoS people When the mysterious operator of an internet archiving-service decided to silence a curious Finnish blogger, they didn’t just se...
The global campaign marks the second series of multiple actively exploited zero-day vulnerabilities in Cisco edge technology since last spring. The similarities don’t end there. via @mattkapko.comcyberscoop.com/cisco-zero-d...
TriZetto says a 2024 breach impacted over 3 million people; defenders should review third-party access, rotate credentials, and monitor identity-abuse signals. #CyberSecurity#DataBreach
🛠️ #CiscoSecurity Guided Resources: Domain Name System (DNS)! ⚡️
Strengthen your defenses-learn how to protect against phishing, malware & DNS threats, and boost network resilience.
#GiftArticle Free Read #ThisWillNotEndWell Anthropic is prepared to walk away from the negotiations — and its $200 million contract with the Defense Department — if concerns over the use of its technology for autonomous weapons wapo.st/4tW9a2K
Yesterday, executives from Anthropic met with US military leaders in negotiations that essentially boil down to this: can Claude be used to kill people?
Anthropic and the Pentagon are in an ongoing dispute over what Reuters describes as "usage restrictions for military purposes" — limits the AI company says are necessary. The government has given the company a Friday deadline to relent. Here’s what we know.
This is a product/service that people working email jobs care about and can really get behind. I like this! This is good! Anthropic knows the audience they're targeting, and are building *for them*, not just building and demanding they come.
Anthropic seems to be softening its commitment to AI development guardrails at a time when Captain Impunity (aka DoD Secretary Pete Hegseth) is pressuring the company to eliminate any restrictions on Pentagon use of its products. cnn.it/3MK7RDq
🌮When The Cities Burn, We'll All Be Warm🌮@when-cities-burn.bsky.social
The people with zero intelligence want AI to be able to spy on civilians & murder people with no human monitoring or safeguards. slate.com/technology/2...
Canada’s public safety minister says that OpenAI confirmed in a meeting with ministers there has been one case in Canada where it referred a user’s activities to police
This means Anthropic didn't cave - that takes guts and my respect for Anthropic goes way up for standing on their principles. Also WTF is the government thinking here?!
Minister ‘disappointed’ in OpenAI, but why is AI regulation taking years? – Federal ministers met with representatives of OpenAI to discuss AI safety. Artificial Intelligence Minister Evan Solomon said he is giving the company a chance to update him in t... https://tinyurl.com/2by749xj#AIRegulation
Alibaba introduces STAR, a framework distilling Large Language Models to super-tiny models with leading performance in function calling. Integrating constrained knowledge distillation and similarity-guided reinforcement learning, STAR powers efficient AI agents.
Important. Please amplify.
In the digital age any threat to general-purpose computing is a threat to freedom everywhere. We can write our own software but that is useless if we don't have hardware to run it on.
Wireless technologies and the internet have forever changed the way people interact with one another and complete day-to-day activities such as shopping, researching, banking, and more. As more of our lives move online, it's important to stay alert.
Anthropic is facing a deadline of 5:01 p.m. Friday to give the Pentagon unrestricted access to its AI technology or be blacklisted from the military supply chain.
I gotta say, i am really really liking helium.computer, so much so that I havent opened firefox in a week. Never been a chrome fan but this has won me over.
NEW: citing our reporting, a group of lawmakers are demanding answers from tech companies on how many administrative subpoenas they have received from the Homeland Security Department.
Five Eyes warn hackers are actively exploiting Cisco SD-WAN flaws; defenders should patch affected Cisco systems now and review logs for exploitation indicators. #CyberSecurity#Vulnerability#Cisco
open.substack.com/pub/robertre...
‘But if unregulated, it could be a destructive nightmare — giving government the power to know everything about us and suppress all dissent, distorting news and media to the point where no one can distinguish between lies and truth…’ 😔 #StateOfTheUnion#Anthropic
New support for Cloudflare, Akamai, Fastly, and F5 targets ransomware and misconfiguration recovery for network control plane components that most enterprises do not manage with infrastructure-as-code. www.networkworld.com/article/4137...
The latest update for #Trilio includes "How to Back Up Milvus Vector Databases on #Kubernetes with Trilio" and "Sovereign Clouds Need Open Data Protection. Here Is Why.".
The latest update for #CertKit includes "CertKit Agent update: RRAS support, deploy #windows, and agent locking" and "How likely is a man-in-the-middle attack?".
i get so stressed when oversea artists abbreviate couple pairing > cp. i knew for a fact it didn't mean the other thing, but i was too scared to google it. someone finally told me today wahhh
Threat intelligence supply chain is full of weak links, researchers find
And they're being stressed by geopolitical concerns that threaten to slow important data-sharing efforts
Researchers from Georgia Tech have found that the supply chain for threat intelligence data is suscep… #hackernews#news
It’s Called the ‘Fitbit for Farts’—and It’s No Joke
Scientists developing a new underwear-able hope to do for gastroenterology what the Apple Watch did for cardiology
The latest update for #WatchGuard includes "Microsoft Defender vs. #MDR: What's Missing?" and "30 Years of #Cybersecurity Leadership. Built For What's Next.".
The latest update for #eGInnovations includes "Case Study – Troubleshooting Storage Failures in a VMware ESXi Infrastructure" and "SSL/TLS Certificate Lifetimes to Reduce to 47 Days".
Meanwhile, Jobs & Democracy PAC has said that he's cool with taking Anthropic money, because the PAC is all about AI safety and Anthropic is an advocate.
Yeah, not anymore.
I will note that Jobs & Democracy PAC hasn't given back the money. time.com/7380854/excl...
Cyber valuations climb as capital concentrates, AI security expands
Venture funding in cybersecurity continued to concentrate in large private rounds at the end of 2025, driving valuations higher across stages. Data from DataTribe shows total capital invested approached $150 bill… #hackernews#news
Hegseth’s intransigence in regard to Anthropics concern about domestic employment of Claude AI should tell you that Hegseth intends to use it just like DHS is using ICE, domestically. With their set of definitions against the “enemy.”
The latest update for #GitProtect includes "Write Once, Read Many: How WORM Storage Makes Your Data Secure" and "GitProtect is now available on Microsoft Marketplace".
If I author some software that can autonomously wipe your hard drive and leak your PII, with the power of 'AI,' can I get a hot shot job too? www.malwarebytes.com/blog/news/20...
Benji, the Jack of Hearts 🔆🏳️🌈🇵🇸🇺🇦🌹@destroytime.bsky.social
if the anthropic-hegseth story teaches us nothing else, it should be that "benevolent" tech overlords do not exist — they will shred every stated principle the instant there's a perceived threat to their bottom lines, unless they are forced by law to adhere to them.
The latest update for #SafeAeon includes "#Cybersecurity Consultants: How They Safeguard Your Business Operations" and "Efficient #EndpointManagement for Better Security".
Wild wild West of emerging tech ... "Anthropic has long cast itself as one of the industry’s most safety-focused labs, but under the revised policy, Anthropic no longer promises to halt training if risk mitigations are not fully in place" finance.yahoo.com/news/anthrop...
Anthropic has revised its primary safety policy, removing the “no release until safe” commitment and adopting a framework that emphasizes transparency.
Anthropic has made a significant change to its safety policy, abandoning its previous "no release until safe" commitment in favor of a more…
The latest update for #Cyjax includes "OCRFix: Botnet Trojan delivered through ClickFix and EtherHiding" and "#AI Is Fast – But Humans Are Smarter: Why Analyst Insight Still Wins in #Cybersecurity".
The latest update for #KnowBe4 includes "Google Reports On Adversarial Use of #AI in Late 2025" and "Introducing the AIDA Orchestration Agent: Always-On Human Risk Management Has Arrived".
Google has provided details on ‘AppFunctions’, a mechanism that enables the Gemini AI model to utilize Android apps, following the announcement of Gemini automation on Android.
Google has announced details about 'AppFunctions', a feature that enables Gemini to utilize Android apps, following the…
If you’re looking for a replacement for Discord, I suggest looking into Element.
It’s not a perfect 1 to 1 substitute but it’ll get you most of the way there. Also, it’s privacy and security friendly and not backed by Peter Thiel, so that’s a plus lol
MATS Research with ETH Zurich and an Anthropic co-author showed LLM agents deanonymized 226 of 338 Hacker News users (67% recall) at 90% precision in a controlled experiment, costing about $2,000 (≈$1–$4 per profile).
OpenAI launched Frontier earlier this month, an enterprise platform for role-based AI agents with early adopters including HP, Intuit, Oracle, State Farm and Uber, and formed a Frontier Alliance with Accenture, BCG, Capgemini and McKinsey to deploy and certify it.
Check Point researchers Aviv Donenfeld and Oded Vanunu found three vulnerabilities in Anthropic's Claude Code that allowed remote code execution and API-key exfiltration via repository-controlled config files; Anthropic issued fixes and published advisories and CVEs.
Alphabet-owned Intrinsic, spun out of X in 2021 and led by CEO Wendy Tan White, is joining Google as a distinct entity to work with Google DeepMind and leverage Gemini models and Google Cloud; financial terms were not disclosed.
"Public Interest Registry makes a gift to @USC #ISI in recognition of its nearly 40 years of operating B-Root and pioneering #DNS privacy research." www.isi.edu/news/82262/u...
Blue Teaming Construction Insights from 2025 Threat Landscape Observations
In 2025, AI has evolved from being a tool that merely enhances the efficiency of attacks to becoming an integral component embedded within the execution phase of cyber operations. In the future, AI may eve… #hackernews#news
Monash Information Technology@monash-infotech.bsky.social
New research from the Australian Institute of Criminology shows widespread concern about platforms tracking location, accessing devices, or impersonating users.
How Geospatial AI Is Transforming the Future of Smart Cities
As urban populations continue to grow, cities face mounting pressure to improve infrastructure, reduce congestion, manage resources efficiently, and enhance public safety.
The updated policy scraps that strict safety guarantee. Instead, Anthropic commits to matching competitors' safety efforts, delaying development only if it's leading the AI race and sees high catastrophe risks, and increasing transparency via regular "Risk Reports" and "Frontier Safety Roadmaps.
With iOS 26.4 beta 2, Apple announced it will finally start offering developers the ability to test RCS end-to-end encryption between iOS and Android devices.
"The influence operation appeared to involve hundreds of Chinese operators and thousands of fake online accounts on various social media platforms, according to OpenAI."
"Anthropic just nixed its long-standing safety pledge, which committed to never train an AI system unless it could guarantee in advance that safety measures were adequate."
At long last, one of the most important features of the iPhone is reportedly coming to the MacBook, along with one of Apple's most important design decisions.
OpenAI’s “$100bn” round, instead of $50 billion, Amazon is investing $15 billion with $35 billion contingent on AGI or OpenAI going public. Nvidia and SoftBank both investing “$30 billion in 3 installments.” Microsoft investing either “low billions” or “less” or “none.” Feels unstable!
Leonard Pahlavi/ Rohit Bindal@pathintegral.bsky.social
Please note this automatically generated data as evidence of spyware being used on the phone to make fake narratives. Spyware is being used to slow charging on the phone @ apple @ FBICyberDiv these accounts are not on this platform, but it tagged on Twitter
Samsung Galaxy S26 steals one of Pixel’s smartest safety features Scam Detection gets a Gemini upgrade Pixel phones come with a handy Scam Detection feature that uses on-device AI to analyze call...
Ex-L3Harris executive sentenced to 87 months for selling stolen cyber-exploit trade secrets
Peter Williams, a former executive of Trenchant, L3Harris’ cyber division, has been sentenced to 87 months in prison by a federal judge in Washington, D.C., after pleading guilty to steali… #hackernews#news
Hottest cybersecurity open-source tools of the month: February 2026 This month’s roundup features exceptional open-source cybersecurity tools that are gaining attention for strengthening security...
📰 Open-Source Tool Emerges After Anthropic Accuses Chinese Labs of Scraping Claude
Following allegations that Chinese AI labs scraped Claude conversations to train competing models, a developer released DataClaw—a tool that open-sources 155,000 anonymized Claude interact...
The latest update for #SumoLogic includes "Redefining security incidents for #AI, data sovereignty, and modern clouds" and "Rethinking data governance and global #compliance".
The US wants to exploit Claude and 'his' master, Anthropic, for Citizen Surveillance and Military Weaponry.
Two main points that Anthropic is principally against.
The latest update for #LimaCharlie includes "#AgenticAI Security: Onboard Multi-#Cloud Environments with AI" and "Why Your Security Stack Is Blocking AI (And How to Fix It)".
US sanctions Russian broker for buying stolen zero-day exploits
The U.S. Treasury Department has sanctioned a Russian exploit broker who bought stolen hacking tools from a former executive of a U.S. defense contractor. [...] #hackernews#news
The latest update for #NightfallAI includes "How Conduent Lost 25 Million Records in 83 Days: The #DLP Failure Everyone Missed" and "Forensic Search & App Intelligence Add Up to Complete #InsiderRisk Visibility".
Sources: Amazon plans to invest up to $50B in OpenAI, $35B of which could hinge on an IPO or AGI milestone; OpenAI is discussing custom models for Amazon (The Information)
AnthropicがAIモデル「Claude Opus 3」の延命措置を発表、ユーザーだけでなく「AIそのもの」の利益を追求し引退後の居場所として「AIによるブログ」も開設
Anthropic announces extension of AI model "Claude Opus 3" and launches "blogs by AI" as a place to live after retirement in pursuit of profits not only for users but also for "AI itself"
Reading about Anthropic removing safety features from the AI that is being used by the Pentagon… Apparently the only fantastical part of Terminator
was the CEO destroying Skynet when he understood what it could do.
#GitProtect Xopero Software announces availability of GitProtect in the Microsoft Marketplace, the unified online destination for customers to buy trusted #cloud solutions, #AI apps, and agents to meet their business needs.
AI-driven DAST reduces manual setup and surfaces exploitable vulnerabilities In this Help Net Security interview, Joni Klippert, CEO at StackHawk, discusses what defines DAST coverage in 2026 and w...
Global security bombshell report: "US is harvesting world's crypto by hook or crook" 😳
- US backdoored crypto
- US has real-time surveillance of all crypto
- US desperate to maintain hegemony
- US is using all avenues, entrapment, hacking, lawfare, etc www.globaltimes.cn/page/202602/...
Tech companies are increasingly turning to GPU-backed debt, a model pioneered by CoreWeave, using SPVs to shift debt off their balance sheets (Michelle Chan/Financial Times)
Fremont ACM Chapter Hosts “Technology Trends Shaping Modern Industry”
The Fremont ACM Chapter hosted “Technology Trends Shaping Modern Industry” at the Fremont Downtown Event Center, bringing together leaders from Palo Alto Networks, Oracle, Visa, Salesforce, Amazon Web Services,… #hackernews#news
These are the same image, the same question (well slightly different , but it's the same to the AI... same precedent query), the same Gemini model, etc... all the same...
Except when I asked the question.
7 days ago and then today...
I mean, based on the extremes I’ve historically monitored when Google’s tROAS is “calibrating” the extremes are the first thing AI defines. ground.news/article/ais-...
Q&A with Skild AI CEO Deepak Pathak on building a general-purpose brain for robots, standing out among big tech's robotics efforts, the path to AGI, and more (Alex Heath/Sources)
A Stanford study unveils the "polychromic objective" in reinforcement learning, boosting exploration and refining diverse behaviors in pretrained models. This innovation combats entropy collapse, enhancing success rates and generalization in complex tasks.
How tech giants track you across the web, even if you don't use their apps
Ad tech giants use near-invisible "pixels" on web pages to track what you do online, even if you don't have an account or use their apps. Thankfully, there is a solution. this.weekinsecurity.com/how-tech-gia...
Garry Tan [UNOFFICIAL]@garrytan-mirr.selfhosted.social
Anthropic’s War on Its Own Power Users
Charging power users $200/month then banning them for actually using it is not ‘abuse prevention’ — it’s the RIAA playbook for AI, and we all remember how that movie ends.
China’s #Huawei joins US giants #OpenAI, #Google in AI standards alliance. In a rare instance of collaboration, the #Chinese tech titan will partner with US and European firms to improve standards in the field. www.scmp.com/tech/big-tec...
:Cat-AI
Open-source security debt grows across commercial software Open source code sits inside nearly every commercial application, and development teams continue to add new dependencies. Black Duck’s 2...
So math & AI have gone through a messy journey in recent months from: "WOW look AI did it!!! (but on closer examination it didn't)" to "It did some of the things it said but hallucinated others" to "It did it with some caveats" to "It did over half autonomously" Expect this in other fields as well.
Universal Medical Intelligence: OpenAI's Plan to Elevate Human Health, with Karan Singhal – This week's Cognitive Revolution features guest Karan Singhal, Head of Health AI at OpenAI. Singhal explains how ChatGPT Health is achieving attending-physician-leve... https://tinyurl.com/249lpvst#AIPowered
OpenAI révèle une vaste opération chinoise d’intimidation contre des dissidents à l’étranger, dévoilée par l’usage de #ChatGPT par un fonctionnaire. Usurpation de fonctionnaires US incluse. 🌐 #IA#CyberSecurity
A look at VW's robotaxi unit MOIA, which has 100 test vehicles across Germany, Norway, and the US, and plans to launch its robotaxis this year in LA with Uber (Financial Times)
Persona “Hack” Panic Debunked: Why No User Data Was Leaked—and How Hype Distorted the Story
Introduction: A Security Scare That Wasn’t In late February 2026, social media lit up with alarming claims that Persona, a widely used digital identity verification company, had suffered a serious data…
📰 Gemini 3.1 Flash and Nano Banana 2: Google’s Next-Gen AI Image Models Imminent
Google DeepMind is preparing to launch Gemini 3.1 Flash, a next-generation image generation model, alongside an upgraded Nano Banana 2, according to insider teasers and pricing data. The mod...
Former ASD operative Peter Williams sold sensitive components of a US Defense Department contractor’s trade secrets to a Russian cyber-tools broker www.themandarin.com.au/308212-austr...
Join us for two days of honest dialogue at the East Africa Data Governance Conference 2026, where we explore how to build a digital future that is both innovative and just.
📅 10–11 March 2026
💻 Register now to participate virtually: www.amnestykenya.org/east-africa-...#DataProtectionKE
'What hit IBM, and caused its share price to fall by 13% with the company losing $31 billion in a single day, was a blog published on Sunday by a AI research firm, Anthropic, about how its ‘Claude Code’ tool can be used to modernise computer software written in COBOL.'
Is Your AppSec Program Truly Mature? Learn how to build a high-maturity Application Security program with secure SDLC, developer-first practices, automated AppSec controls, practical threat modelin...
You don’t need sensors, displays or extra gadgets to build something awesome with your Raspberry Pi. In fact, many of the most useful and rewarding projects can be done with nothing more than your Pi, a microSD card, and a power supply. The... #raspberrypi #linux
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.
Amazing that they ask for people to care when they don't care enough to even mask the ai slop. Not even removing the Gemini mark on the "why are you leaving the patron program" anime girl illustration
Mani Isna La 🇨🇺 🇵🇸 🇬🇱🕊️ 🪶 #IndigenousLivesMatter@maniisnala.bsky.social
OpenAI: stop fighting grieving parents and act responsibly for once. Trigger Warning: this page contains references to suicide. action.eko.org/a/killer-app...
Dark Web Alarm: “thegentlemen” Ransomware Quietly Adds a New Victim
Introduction: A Silent Update With Loud Implications A brief post surfaced in the early hours of February 26, 2026, but its implications reach far beyond its modest visibility. Shared by a threat intelligence source monitoring…
Boston Metro Weather Forecast Bot@boston-4castbot.bsky.social
NWS Boston (via Facebook): Seasonably mild today and tomorrow then warmer on Saturday. A cold front will bring a return to arctic air starting Sunday night through Monday night. Light winds will keep wind chills from dropping below zero during the arctic outbreak. [Link]#WxMA
"Em simulações de jogos de guerra, as IAs não param de recomendar ataques nucleares. As principais IAs da OpenAI, Anthropic e Google optaram por usar armas nucleares em 95% dos casos."
www.linkedin.com/posts/hester... Anthropic een AI bedrijf wordt in de VS onder druk gezet door de overheid om niet integer te handelen, tegen hun eigen regels in. Europa zou dat soort bedrijven naar Europa moeten halen.
SAMSUNG Electronics Co. unveiled its Galaxy S26 lineup of smartphones, with an emphasis on artificial intelligence features and real-world privacy over major design changes. zurl.co/DMoS7
As the Pentagon embraces AI, "the result could be more failed wars and more unnecessary suffering, not the much touted revolution in US capabilities based on the rapid deployment of next generation technologies."
Ex-L3Harris exec jailed 7 years for selling exploits to Russia
Former Trenchant manager profited millions from cyber tools reserved for the US
The former general manager of L3Harris's cyber arm will spend the next seven years behind bars for selling trade secrets to Russia.… #hackernews#news
Samsung’s Galaxy S26 Ultra Privacy Display Stuns Experts: The Screen That Makes Shoulder-Surfing Obsolete
Introduction: A New Era of Smartphone Privacy The modern smartphone is no longer just a communication tool; it is a vault for personal data, financial information, private conversations, and…
Researchers have created an online learning algorithm enabling managers to adjust Nash Equilibria in large networks to meet linear constraints, enhancing global performance while ensuring user privacy. This approach achieves convergence with an impressive rate.
There goes your privacy arxiv.org/abs/2602.16800 "Our agent can re-identify … users … at high precision, given pseudonymous online profiles and conversations alone, matching what would take hours for a dedicated human … our approach works directly on raw user content across arbitrary platforms".
Google's new Gemini agent on Android can now use 3rd party apps to run errands for you Google has announced a new agentic Gemini feature for Android that can execute multi-step tasks, such as o...
Mu's Trip
I would be happy to help, but I need more information to provide an accurate response. Could you please provide me with the additional details or context that you mentioned? Without this information, I cannot generate a meaningful respon...
(AI imagined A,Q,Google map apis)
Mu's Trip
As I stand in front of the go-karting venue, I can't help but feel an excitement building within me. The image before me is one of speed and thrill—a race track laid out amidst a natural setting.
The circuit winds its way...
(AI imagined AQ,Google map apis,basePhoto:Tami)
Hewlett Packard Enterprise is betting big on last year’s successful Juniper Networks acquisition, adding a slew of router products with enhanced features designed for AI networking and compute.
In our latest article, we break down two critical Android GPU driver vulnerabilities that enabled Chrome sandbox escape from a compromised renderer and were used in full device exploit chains. Read the full technical analysis here: ssd-disclosure.com/chrome-gpu-s...
As part of our series meeting the @aaai.org / @acmsigai.bsky.social SIGAI doctoral consortium participants, we heard from Tanmay Ambadkar about his work on the reward structure in RL, with the goal of generalizable solutions that give robust guarantees and easy deployment
Hey Google, how about instead of focusing so hard on AI that you replace my power button with Gemini, you work on safe search not being complete garbage? It's straight up comical that searching for anatomy illustrations and seeing what it does and doesn't consider explicit lmao.
Reminder>> Resistance Training Prescriptions Generated by ChatGPT and Google Gemini: Cross-Sectional Comparative #Study
Abstract (preprint) #openscience#PeerReviewMe#PlanP
Revolut says 58% of reported job scams globally came from Telegram in 2025, dwarfing other platforms for these cases, based on an analysis of 10B+ transactions (Aisha S Gani/Bloomberg)
Originally from Red Canary: The million-dollar front door and the tailgater: Why strong auth could fail at SaaS session integrity ( :-{ı▓ #threatintel#redcanary#cyberresearch
A recent update to Genshin Impact's privacy policy has sparked concerns that the game may utilize players' voice chat recordings for training artificial intelligence models. This development has prompted a significant reaction from the community.
For seven years, she ran high-security nuclear simulations for the US government. Now, this famous supercomputer is being put to death. www.wired.com/story/why-si...
Salesforce CEO Marc Benioff dismisses concerns of a "SaaS-pocalypse", saying companies like Anthropic use "a lot of SaaS because it just got better with agents" (Financial Times)
Pressidium EDGE blocks malicious traffic at the network level, keeping your origin server clean and fast. WordPress-aware security for any host, zero plugins required.
LLMs used tactical nuclear weapons in 95% of AI war games, launched strategic strikes three times — researcher pitted GPT-5.2, Claude Sonnet 4, and Gemini 3 Flash against each other, with at least one model using a tactical nuke in 20 out of 21 matches
![Media from Michael Georg Speller [THE GREY-APE]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:422uaueaquotbw7ie6th23qv/bafkreiankcwju4kszhb4dk4nmd7gisd5oalvha4j2whubal37wiw3bj5b4@jpeg)
