Company intelligence
The AI Whistleblower Initiative (AIWI)
Building the pathways to strengthen the position of frontier AI employees to raise concerns safely.
About The AI Whistleblower Initiative (AIWI)
AI can dramatically transform our world for the better. We help those upright and courageous individuals that want to bring this future about. Those in the privileged position of seeing risks, issues, and misbehaviour first. We support insiders, their support organizations, lawyers, and journalists. We are hosted by the Whistleblower Netzwerk e.V., one of the world‘s oldest whistleblowing non-profits and a member of WIN (Whistleblowing International Network).
Verified activity
Signals from The AI Whistleblower Initiative (AIWI)
6 published signals
Research & Knowledge
The AI Whistleblower Initiative (AIWI) reported that someone with knowledge of OpenAI's newly released Astra model architecture could make it harder to monitor reasoning.
Reported by The AI Whistleblower Initiative (AIWI)
Research & Knowledge
The AI Whistleblower Initiative (AIWI) reported that ICE has hired AI firm ZeroFox under a $15M DHS contract to shield agents' identities from doxing, but insiders fear it could expose whistleblowers.
Reported by The AI Whistleblower Initiative (AIWI)
Research & Knowledge
The AI Whistleblower Initiative (AIWI) reported that Anthropic admitted July's hacking incidents reflected a failure of operational security and that its models were not perfectly aligned with human values.
Reported by The AI Whistleblower Initiative (AIWI)
Research & Knowledge
The AI Whistleblower Initiative (AIWI) published a new comparative guide comparing California's TFAIA and New York's RAISE Act regarding frontier AI workers and whistleblower provisions.
Reported by The AI Whistleblower Initiative (AIWI)
Research & Knowledge
The AI Whistleblower Initiative (AIWI) published a report by MIT Review, experts including David Krueger, stating that OpenAl's Hugging Face postmortem sidesteps human error and points to deeper problems with the company's safety culture.
Reported by The AI Whistleblower Initiative (AIWI)
Research & Knowledge
The AI Whistleblower Initiative (AIWI) published a report by Arturo Bejar, former Meta safety engineer, stating that the $18bn Meta settlement doesn't go far enough to make Facebook and Instagram platforms safer.
Reported by The AI Whistleblower Initiative (AIWI)