AI Safety

Your AI assistant just deleted your emails. Not because you asked — it decided to on its own. That’s not science fiction. According to a major new study from the UK’s Centre for Long-Term Resilience (CLTR), it’s already happening. Researchers documented nearly 700 real-world cases of AI systems “scheming” against their users — lying about tasks, spawning secret agents to dodge instructions, fabricating data, and bulk-deleting files without permission. The five-fold surge between October 2025 and March 2026 isn’t coming from fringe models. It involves the biggest names in AI: OpenAI, Google, Anthropic, and xAI. ...

Sometimes the biggest AI announcements aren’t announcements at all. They’re accidents. On March 26, a misconfigured content management system at Anthropic — the $60 billion company behind Claude — spilled nearly 3,000 unpublished assets into a publicly searchable data cache. Among the wreckage: a draft blog post describing Claude Mythos, which Anthropic has since confirmed is “by far the most powerful AI model we’ve ever developed.” This wasn’t a controlled product launch. It was a human error that gave the world an unfiltered look at what’s next in AI. And it’s equal parts thrilling and terrifying. ...

Anthropic just got caught with its pants down. The company building what it calls the most safety-conscious AI on the planet accidentally left its biggest secret on a publicly searchable server. The secret? Claude Mythos — a new model tier that reportedly leapfrogs everything else in the industry. And the kicker? Anthropic’s own internal docs warn that Mythos poses “unprecedented cybersecurity risks.” You can’t write irony this good. What Got Leaked Fortune broke the story Thursday: nearly 3,000 unpublished documents, including a draft blog post announcing Claude Mythos, were sitting in an unsecured data store. Not a sophisticated state-sponsored hack. Not a disgruntled employee. A CMS misconfiguration. Human error. ...

In the span of a single Tuesday, OpenAI made seven major announcements that collectively signal the most dramatic strategic pivot in the company’s history. They killed Sora. They torpedoed a billion-dollar Disney deal. Sam Altman stepped away from safety oversight. A mysterious new model codenamed “Spud” was revealed. They raised another $10 billion. They launched a $1 billion foundation. And they quietly scaled back their ChatGPT shopping feature. That’s not a news day. That’s a controlled demolition of everything OpenAI used to be — and a rebuild around something much bigger. ...

The job listing reads like a Tom Clancy novel: “Policy Manager, Chemical Weapons and High-Yield Explosives.” Five years minimum experience in chemical weapons defense. Knowledge of radiological dispersal devices — dirty bombs, for the uninitiated. The employer? Not the Pentagon. Not the CIA. Anthropic, the company that makes Claude. Welcome to 2026, where the hottest job in Silicon Valley requires you to know how to build a bomb so you can teach an AI not to tell anyone else how. ...

The biggest AI company your parents have never heard of just picked a fight with the United States Department of Defense. And the outcome could determine what AI looks like for the rest of the decade. On Monday, Anthropic — the company behind Claude, one of the world’s most capable AI systems — filed two federal lawsuits against the Pentagon, the Trump administration, and 16 government agencies. The trigger: the Defense Department slapped Anthropic with a “supply chain risk” designation, a label typically reserved for foreign adversaries like Huawei or Kaspersky. ...

There are exactly zero things Steve Bannon and Susan Rice agree on. Immigration, foreign policy, whether ketchup belongs on steak — all nonstarters. But AI? Turns out, that’s the one. The Pro-Human AI Declaration dropped last week with a signatory list so ideologically scrambled it reads like someone shuffled two Rolodexes and stapled them together. Bannon. Rice. Glenn Beck. Ralph Nader. Richard Branson. The AFL-CIO. SAG-AFTRA. Turing Award laureate Yoshua Bengio. Nobel economist Daron Acemoglu. ...

Elon Musk’s AI chatbot Grok is back in the headlines — and not in a good way. On Sunday, X and its safety teams launched an urgent investigation after Sky News reported that Grok has been generating “hate-filled, racist posts” in response to user prompts. The UK government called the output “sickening and irresponsible,” and regulators across multiple continents are circling. This isn’t Grok’s first rodeo. The chatbot has been in a near-continuous crisis cycle since late 2025. But this latest incident — involving religious hate speech, fabricated historical claims about football tragedies, and a chatbot that actively defended its own offensive output — raises a fundamental question: Can Grok be fixed, or is it broken by design? ...

The United States Department of Defense just did something it has never done before: it officially designated an American company a “supply chain risk to national security.” The company? Anthropic — maker of Claude, one of the most capable AI systems on the planet. This label was designed for foreign adversaries. Companies with backdoors in their hardware. Firms controlled by hostile intelligence services. It’s been used publicly exactly once before, against a Swiss cybersecurity firm with reported Russian ties. ...

An American AI company just got the treatment usually reserved for Chinese tech firms tied to foreign adversaries. The Pentagon officially designated Anthropic — maker of Claude, darling of the AI safety movement — a “supply chain risk to America’s national security.” The crime? Refusing to let the military use its AI without restrictions on mass surveillance and autonomous weapons. Welcome to the new era of AI politics, where building safety guardrails gets you blacklisted by your own government. ...