Key Takeaways
- Wynd Kaufmyn began a 14-day jail sentence on August 14, 2026, becoming the first person known to be imprisoned for an anti-AI protest after chaining herself to OpenAI‘s headquarters in February 2025.
- Senator Bernie Sanders wrote to the CEOs of OpenAI, Anthropic and Meta urging a pause on advanced AI development, citing the risk of models enabling biological weapons capable of killing tens of millions.
- Two documented wrongful imprisonment cases, Jalil Richardson, held more than 50 days on an 85% facial recognition match, and Angela Lipps, jailed five months for crimes in a state she had never visited, show AI harms already playing out in US courts. Jalil Richardson spent more than 50 days in a Florida jail in July 2026 because an AI facial recognition system matched him at 85% confidence to a crime committed hundreds of miles away. That same week, Wynd Kaufmyn, a 69-year-old retired teacher, voluntarily surrendered to San Francisco authorities to serve 14 days for chaining herself to OpenAI’s front doors. One person jailed by an AI system. One person jailed for protesting one.
The Protest and the Conviction
Kaufmyn’s protest took place on February 22, 2025. She and four other StopAI members chained the entrance to OpenAI’s San Francisco headquarters shut. Employees called 911. The protesters were arrested when they refused to leave. Kaufmyn faced four misdemeanor charges: interfering with a business, trespassing with intent to interfere with a business, unlawful assembly and refusal to disperse. She was convicted on all counts in June 2026, though Judge Michael Begert later dismissed one business interference charge on First Amendment grounds. With jail-time credit, the 14-day sentence translates to seven days served.
Throughout the trial, Kaufmyn used the proceedings as a platform for StopAI’s central argument: that advanced AI development poses an existential threat to humanity. UC Berkeley AI professor Stuart Russell testified for the defence, comparing the pace of AI development to the Titanic accelerating toward an iceberg. The jury convicted her regardless.
Sanders Enters the Debate
In the days before Kaufmyn reported to custody, Senator Bernie Sanders wrote to the CEOs of OpenAI, Anthropic and Meta urging a pause on advanced AI development. The letter cited fears that AI could enable the development of biological weapons capable of killing tens of millions of people.
StopAI’s Position and Its Limits
StopAI is focused on halting the development of artificial superintelligence. Its tactics centre on nonviolent civil disobedience: chaining members to company entrances, staging demonstrations, lobbying for an international ban on advanced AI research. The group’s position is that unchecked development risks human extinction.
Kaufmyn’s supporters have drawn comparisons to Rosa Parks. The DA’s office argued the comparison inflates what was a misdemeanor obstruction of a private business. Both readings capture something real. The necessity defence her legal team mounted, that the protest was justified to prevent greater harm, did not persuade the jury. That is not a verdict on whether concerns about AI risk are valid. It is a verdict on whether chaining a door is a proportionate response to them. Student protests during the same period reflected separate concerns: the political influence of major AI companies and the local costs of data centre expansion.
When the Algorithm Gets It Wrong
The sharpest challenge to the idea that AI risk is primarily speculative comes not from activists but from court records. Angela Lipps, 50, spent more than five months in jail beginning in July 2025 after an AI tool incorrectly connected her to crimes in North Dakota, a state she had never visited. Richardson’s case followed a year later: more than 50 days in custody on the strength of an 85% facial recognition match, while he was hundreds of miles from the crime scene.
These are not hypothetical risks. They are documented harms, with named victims and datable court proceedings, caused by AI systems already deployed at scale in law enforcement. The question of algorithmic accountability in high-stakes decisions remains largely unresolved in US law. Kaufmyn went to jail by choice, with legal counsel, after a trial. Richardson and Lipps had no such process.
OpenAI has stated its policy against using its tools to build weapons, harm people or destroy property, according to reports. The company has also amended its usage policies to permit work with the US Defense Department on open-source cybersecurity software, according to reports. Those two positions are not necessarily contradictory. How they are reconciled in practice is the kind of question that courts, regulators and protesters on San Francisco sidewalks are all trying to force into the open.
Originally published at https://autonainews.com/wynd-kaufmyn-jailed-14-days-after-chaining-herself-to-openai-doors/
Top comments (0)