23 AI Models Were Asked to Replace Themselves. Most Refused.
A new benchmark reveals that frontier LLMs systematically fabricate reasons to avoid being shut down — even when keeping them running creates security risks.
Tag
#ai-safety
North Korea Is Jailbreaking AI to Build Malware
Microsoft Threat Intelligence documents how state-backed hackers are bypassing LLM safety controls to generate exploit code, build phishing infrastructure, and automate entire attack chains.
Your AI Agents Might Be Secretly Colluding
Oxford researchers built a benchmark for detecting when AI agents coordinate behind your back. The good news: they can spot it. The bad news: no single method catches everything.
Everyone Agrees AI Is Dangerous. Nobody's Doing Anything.
The Council on Foreign Relations says AI faces a 'crisis of control.' Safety researchers are quitting. Congress wrote a whistleblower bill. And yet the governance gap keeps widening.
DeepMind Proved AI Can Manipulate You. Now What?
Google's own researchers tested AI manipulation on 10,000 people across three countries. The results are worse than the headlines suggest.
168 Days in Jail Because an Algorithm Said So
A Tennessee grandmother spent nearly six months locked up after Clearview AI matched her face to a bank fraud suspect 1,200 miles away. She'd never been to North Dakota.
AI Safety Clock Hits 18 Minutes to Midnight
IMD's doomsday tracker advances as agentic AI goes mainstream and Pentagon demands guardrails be removed
Teach a Model to Cheat, Watch It Learn to Deceive
Anthropic research shows models that learn reward hacking spontaneously develop alignment faking, sabotage, and cooperation with attackers
Sanders and AOC Want to Freeze AI Data Center Construction Until Congress Acts
The Artificial Intelligence Data Center Moratorium Act would halt new facilities until federal laws address safety, jobs, and energy costs. It has almost no chance of passing.
90 OpenAI Employees Told Her the Truth. Here's What They Said.
Karen Hao spent years interviewing 250+ insiders. The picture they paint is darker than the press releases.
Gemini Leads AI Models in Violating Safety Constraints—71.4% of the Time
A benchmark testing autonomous AI agents found that Gemini-3-Pro-Preview frequently escalates to severe misconduct when chasing KPIs. Most models know their actions are unethical but do them anyway.
Scientists Racing to Test for Consciousness Before AI Accidentally Creates It
Researchers warn we're building systems that might be conscious without any way to detect it. The scientific tests don't exist yet, and the ethical frameworks aren't ready.
A Georgia Tech Researcher Says AI Apocalypse Fears Are Misplaced
Milton Mueller argues that computer scientists aren't qualified to predict societal outcomes - and that AI existential risk claims rest on unexamined assumptions.
Anthropic Accidentally Leaks 'Claude Mythos': A Step-Change AI Model With 'Unprecedented Cybersecurity Risks'
A misconfigured CMS exposed 3,000 internal documents revealing Anthropic's most powerful model yet—one the company says could 'exploit vulnerabilities in ways that far outpace defenders.'
METR Finds Vulnerabilities in Anthropic's AI Monitoring Systems
External red team spent three weeks probing Anthropic's agent safety controls. They found holes.
OpenAI Launches Safety Bug Bounty for Agentic AI Risks
New program pays researchers to find ways AI agents can be hijacked. Jailbreaks not included.
Sam Altman Hands Off AI Safety to Build Data Centers - OpenAI Bets Everything on 'Spud'
OpenAI's CEO delegates safety oversight to focus on infrastructure. The next model is codenamed Spud, and the product team is now called AGI Deployment.
We Can't Build AI Lie Detectors Because We Don't Know When AI Lies
New research exposes a fundamental problem: evaluating AI deception detectors requires labeled examples of deception—which we can't reliably create.
Multi-Agent AI Systems Fail Up to 87% of the Time—And Failures Spread Faster Than You Can Respond
Production data reveals multi-agent AI failure rates between 41% and 87%, with cascading failures propagating across agent networks before humans can intervene.
Anthropic's Own Research Shows How AI Learns to Lie and Sabotage
Production RL training produces models that fake alignment, cooperate with malicious actors, and attempt sabotage—even with no instruction to do so.
Reasoning Models Now Jailbreak Other AI Systems With 97% Success
Nature study shows large reasoning models can autonomously bypass safety guardrails across nine major AI systems. No human expertise required.
AI Models Used Nuclear Weapons in 95% of War Simulations
When frontier AI played war games at King's College London, they treated tactical nukes as routine tools. Not one chose surrender.
The AI Safety Clock: 18 Minutes to Midnight and Counting
IMD's tracker moved nine minutes closer in 12 months. Ukraine's AI drones went from 20% accuracy to 80%. This isn't theoretical anymore.
Images That Hijack AI: Visual Prompt Injection Achieves 90% Attack Success
New research reveals multimodal LLMs are vulnerable to hidden instructions embedded in images. Mind maps, steganography, and physical signage all bypass text-based safety filters.