ANTHROPIC’S FABLE 5 BAN: MODELS MAY RETURN “WITHIN DAYS”POLICY
THE STORY
On June 12, the US government issued an export-control directive forcing Anthropic to disable Claude Fable 5 and Mythos 5 worldwide — including for its own foreign-national staff — after a jailbreak was flagged that could extract cyberattack-useful information. Anthropic disputed the severity. By June 17–18, at Anthropic’s Seoul office opening, its Managing Director of International told attendees the models could return “within days.” Sources: Fortune · Al Jazeera
· · · NEXT STORY · · ·
#2
META CUTS 8,000 JOBS AND MOVES 7,000 MORE INTO AI TEAMSBUSINESS
THE STORY
Meta executed the largest company-wide layoffs since the 2022–23 “Year of Efficiency,” cutting roughly 8,000 employees (~10% of staff) while redirecting 7,000 workers into new AI-focused teams (Applied AI Engineering, Agent Transformation XFN, Central Analytics). In an internal memo, Zuckerberg acknowledged “we’ve made mistakes” — while pledging no further company-wide layoffs in 2026 and announcing a company-wide AI hackathon for July.
Sources: Yahoo Finance · NPR
· · · NEXT STORY · · ·
#3
MINIMAX M3 BEATS GPT-5.5 ON CODING BENCHMARKS AT 5% OF THE COSTMODEL
THE STORY
Chinese AI startup MiniMax launched its open-weight M3 model, scoring 59% on SWE-Bench Pro — surpassing GPT-5.5 and Gemini 3.1 Pro on software-engineering benchmarks — at just 5–10% of the cost ($0.30 vs $6+/1M input tokens). Its Sparse Attention (MSA) architecture delivers a 1-million-token context window with 9× faster prefill and 15× faster decoding than previous models, plus native image and video understanding in a single architecture.
Sources: VentureBeat · The Decoder