The order came Friday evening. At 5:21pm ET on June 12, a US export-control directive forced Anthropic’s Fable 5 shutdown, suspending access to both Fable 5 and Mythos 5 worldwide just three days after their June 9 launch. The government cited national security and a claimed jailbreak method. Anthropic says it cannot filter foreign nationals in real time, so it disabled both models for every customer globally. It calls a recall disproportionate for models deployed to hundreds of millions. Its other models stay live.
The scrutiny didn’t stay with one lab. A coalition of state attorneys general opened a probe into OpenAI, with New York’s OpenAI subpoena served Friday across five areas: advertising, engagement and retention, model sycophancy, health-data handling, and treatment of minors and seniors. It lands just four days after OpenAI filed confidentially for an IPO on June 8. OpenAI says it takes the concerns seriously, intends to engage constructively, and pointed to its age-prediction tools and parental controls. The pressure is widening from export law to consumer protection.
The credibility question surfaced inside the enterprise, too. Over the weekend, KPMG’s withdrawn AI report came down after four named organizations — UBS, the UK’s NHS, Swiss Federal Railways, and Transport for London — said its descriptions of their work were inaccurate. Detection firm GPTZero flagged the passages as AI-generated fabrications in a report about agentic AI. KPMG, which is embedding Claude across roughly 276,000 employees, says it expects human oversight to validate content. Shipping a frontier model is one problem; standing behind its output is another.
That’s the signal.
— The PulseMark Team
Get the Daily Pulse
Sharp analysis on what's actually moving in AI. No hype, no filler, no weekly digest.