3 months ago
Individuals with past and contiguous roles astatine OpenAI and Google DeepMind called for the extortion of critics and whistleblowers connected June 4.
Authors of an unfastened missive urged AI companies not to participate agreements that artifact disapproval oregon retaliate against disapproval by hindering economical benefits.
Furthermore, they stated that companies should make a civilization of “open criticism” portion protecting commercialized secrets and intelligence property.
The authors asked companies to make protections for existent and erstwhile employees wherever existing hazard reporting processes person failed. They wrote:
“Ordinary whistleblower protections are insufficient due to the fact that they absorption connected amerciable activity, whereas galore of the risks we are acrophobic astir are not yet regulated.”
Finally, the authors said that AI firms should make procedures for employees to rise risk-related concerns anonymously. Such procedures should let individuals to rise their concerns to institution boards and outer regulators and organizations alike.
Personal concerns
The letter’s thirteen authors described themselves arsenic existent and erstwhile employees astatine “frontier AI companies.” The radical includes 11 past and contiguous members of OpenAI, positive 1 past Google DeepMind subordinate and 1 contiguous DeepMind member, formerly astatine Anthropic.
They described idiosyncratic concerns, stating:
“Some of america reasonably fearfulness assorted forms of retaliation, fixed the past of specified cases crossed the industry.”
The authors highlighted assorted AI risks, specified arsenic inequality, manipulation, misinformation, nonaccomplishment of power of autonomous AI, and imaginable quality extinction.
They said that AI companies, on with governments and experts, person acknowledged risks. Unfortunately, companies person “strong fiscal incentives” to debar oversight and small work to stock backstage accusation astir their systems’ capabilities voluntarily.
The authors different asserted their content successful the benefits of AI.
Earlier 2023 letter
The petition follows an April 2023 unfastened missive titled “Pause Giant AI Experiments,” which likewise highlighted risks astir AI. The earlier missive gained signatures from manufacture leaders specified arsenic Tesla CEO and X president Elon Musk and Apple co-founder Steve Wozniak.
The 2023 missive urged companies to intermission AI experiments for six months truthful that policymakers could make legal, safety, and different frameworks.
The station OpenAI, DeepMind insiders request AI whistleblower protections appeared archetypal connected CryptoSlate.
English (US) 