House ranking Democrats ask five labs to list every model breach of internal, internet or third party networks

Sourcing note: this item rests on a single reporting chain, Punchbowl reporting carried by Bloomberg Law on October 28, 2026; no independent second outlet has reported the letter yet, and this item will be updated when one does. Reps. Frank Pallone, Zoe Lofgren, and Jamie Raskin, the ranking members on key House committees with AI oversight, wrote to OpenAI, Anthropic, Google DeepMind, Meta, and SpaceXAI asking for a list of misalignment incidents that resulted in successful or attempted unauthorized access of internal networks, internet usage, or third party networks by AI models, plus each company's processes to identify and respond to such incidents.

As minority members, the three lawmakers cannot compel answers. The letter is a new letter, distinct from the September 29 and 30, 2026 congressional letters on the same theme.

Sources: Bloomberg Law (Punchbowl)