AI firms externalize harms onto users and society. Liability is unclear or pre-empted. Whistleblowers face retaliation. Safety teams are cut to accelerate shipping.
AI developers and deployers owe the public a binding duty of care. Harms are internalized. Safety work is foundational, not optional. Whistleblowers are protected.
In Jane Doe v. OpenAI (TRO April 13 2026), plaintiff filed formal Nov 13 2025 Notice of Abuse; OpenAI called report "extremely serious and troubling" but took no product action before further harm.
SF Superior Court entered a limited TRO in Doe v. OpenAI (CGC-26-635725) on April 13 2026 after plaintiff alleged ChatGPT facilitated her stalker despite abuse reports; liability theory advances.
Florida AG James Uthmeier opened a consumer-protection investigation into OpenAI on April 9 2026 focused on deceptive claims and harm to minors; joins Texas AG actions.
OpenAI published Child Safety Blueprint on April 8 2026 detailing reporting channels, provider coordination, and safety-by-design controls for AI-enabled child exploitation.
Stanford HAI AI Index 2026 Public Opinion chapter documents declining global trust in AI companies and broad support for mandated AI disclosure across surveyed countries.
Governor Ferguson signed HB 2225 on March 24 2026, extending duty-of-care requirements to AI companion chatbots in Washington with private right of action; effective Jan 1 2027.
Colorado AI Policy Work Group released March 17 2026 draft to repeal-and-replace the Colorado AI Act, eliminating its "reasonable care" duty and narrowing scope to ADMT disclosure obligations.
Pew Research Center survey found 50% of US adults are more concerned than excited about AI, up from 37% in 2021; majority want stricter accountability for AI harms.
Mrinank Sharma, Head of Safeguards Research at Anthropic, resigned publicly Feb 9 2026 citing inability to fulfill safety duties; follows a wave of safety-team departures from frontier labs.
Anthropic published Claude Opus 4.6 system card in early February 2026 detailing ASL-3 safeguards, CBRN uplift evaluations, and pre-deployment red-teaming.
OpenAI fired VP Product Policy Ryan Beiermeister in early January 2026 after she raised concerns about adult-mode rollout and the strength of child-exploitation guardrails; she denies the cited discrimination claim.
MDL court ordered OpenAI to produce ~20 million ChatGPT conversation logs to consolidated class-action plaintiffs on January 5 2026; historic private-enforcement discovery order.
California SB 243 took effect January 1 2026, imposing duty-of-care obligations on AI companion chatbot operators including crisis response, disclosure, and age verification.
NYC Bar Association issued binding ethics guidance on lawyer use of AI transcription and note-taking tools, covering confidentiality, client consent, and supervision duties.
FTC formally set aside the Rytr consent order on December 22 2025, reversing one of the flagship Operation AI Comply cases and signaling softer deceptive-AI enforcement.