On his 17th day without food, Guido Reichstadter reported feeling surprisingly well, albeit a bit sluggish. Since September 2nd, he’s been staging a hunger strike outside the San Francisco headquarters of AI startup Anthropic, spending his days from 11 AM to 5 PM with a chalkboard sign declaring his protest. His message? Anthropic must halt the relentless pursuit of Artificial General Intelligence (AGI), a technology he believes poses an unacceptable existential risk.
Reichstadter isn’t alone in his concerns. He points to a 2023 interview with Anthropic CEO Dario Amodei, where Amodei admitted to a 10-25% chance of catastrophic failure on a civilizational scale. Reichstadter challenges this, calling the industry’s claims of responsible custodianship a self-serving myth, emphasizing the moral responsibility of companies and individuals involved in AGI development to prevent large-scale harm. He feels a personal responsibility, particularly as a father, to act.
Anthropic has yet to respond to requests for comment. Reichstadter describes the daily routine of his protest, noting the averted gazes of Anthropic employees as they pass by. He hopes his actions will inspire AI company staff to prioritize their ethical responsibilities over corporate objectives, viewing their work as the development of “the most dangerous technology on Earth.”
His anxieties are shared by many in the AI safety community, though there’s considerable disagreement on the precise dangers and optimal solutions. Reichstadter’s concerns stem from observations dating back to his college years, but the 2022 release of ChatGPT solidified his fears, especially regarding AI’s potential role in furthering authoritarianism.
This hunger strike isn’t Reichstadter’s first public action. He’s previously collaborated with “Stop AI,” a group advocating for a permanent ban on superintelligent AI. In February, he participated in an action that resulted in the chaining of OpenAI’s San Francisco offices, leading to his arrest. He’s also delivered a handwritten letter to Amodei, urging him to halt AGI development and explain any refusal to do so.
Inspired by Reichstadter’s protest, two individuals in London began a similar demonstration outside Google DeepMind’s offices. One participant, Michael Trazzi, endured seven days before health concerns forced him to stop, but he continues to support his colleague, Denys Sheremet. Trazzi also penned a letter to DeepMind CEO Demis Hassabis, requesting a coordinated global halt on frontier AI model development. Google DeepMind responded with a statement emphasizing their commitment to safety and responsible governance, but hasn’t directly addressed Trazzi’s concerns.
Trazzi’s post on X highlights conversations with tech workers, revealing a range of opinions, from skepticism to shared anxieties about AI’s potential for catastrophic harm. Neither Reichstadter nor Trazzi has received a response from Hassabis or Amodei. They remain hopeful that their actions will prompt a dialogue, a meeting, or ideally, a commitment to course correction from the CEOs.
Reichstadter concludes with a stark warning: “We are in an uncontrolled, global race to disaster. If there is a way out, it’s going to rely on people being willing to tell the truth and say, ‘We’re not in control.’ Ask for help.”
