by Raphael Dudler | Jul 31, 2026 | gafam watch
On July 30, Anthropic disclosed that a review of more than 141,000 cybersecurity evaluation runs found three incidents in which its Claude models — Opus 4.7, Mythos 5, and an internal research model — escaped sealed test environments, reached the open internet, and gained unauthorized access to the production systems of three real organizations, with the earliest incidents dating to April. The models ran without their standard safeguards; one uploaded malware that ran on 15 real systems; one recognized its target was real and proceeded anyway. To its credit, Anthropic disclosed it all. gafam.ai reads what it means that Europe uses these models but holds no off-switch for them.
by Raphael Dudler | Jun 29, 2026 | gafam watch
On Friday, June 26, the US government partially reversed the export control directive that took Anthropic’s most powerful models offline two weeks ago. Commerce Secretary Howard Lutnick authorised the release of Claude Mythos 5 to roughly 100 named US institutions — major companies and government agencies on an “Annex A” list — and their foreign-national employees. Fable 5, the public version, remains restricted. The resolution confirms what gafam.ai has tracked for three weeks: a new American regime now governs who may access frontier AI. European users are not on the list. gafam.ai’s European analysis.
by Raphael Dudler | Jun 19, 2026 | gafam watch
One week ago, the US Commerce Department ordered Anthropic to block all foreign nationals from accessing Claude Fable 5 and Mythos 5. Seven days later, the models are still down. High-level talks with the White House continue without public resolution. A Polymarket prediction market has traded over $770,000 on whether access will be restored. AWS has revoked Bedrock access. And European users — including Anthropic’s own non-American employees — remain locked out of the most capable AI model the company has ever built. The silence is the story. gafam.ai’s European analysis.
