Internal Memo Reveals Friction Between AI Rivals
Anthropic Chief Executive Dario Amodei has launched a scathing critique against OpenAI and its CEO, Sam Altman, regarding their recent strategic partnership with the U.S. Department of Defense (DoD). In a leaked staff memo reported by The Information, Amodei characterized OpenAI’s justifications for the deal as "safety theater" and accused the competitor of dishonesty regarding safety protocols.
The Breaking Point in Negotiations
The conflict stems from failed negotiations between Anthropic and the DoD. Despite holding an existing $200 million contract with the military, Anthropic walked away from a new agreement because the DoD insisted on language permitting the AI models for "any lawful use." Amodei stated that Anthropic refused to comply without explicit guarantees that the technology would not be utilized for domestic mass surveillance or autonomous weapons systems.
OpenAI subsequently secured the deal, asserting that their contract includes safeguards against these specific red lines. However, Amodei disputed this claim in his correspondence, alleging that OpenAI prioritized placating their workforce over actually preventing potential abuses. He argued that Altman is falsely positioning himself as a dealmaker while obscuring the reality of the contract's broad permissions.
Legal Definitions vs. Ethical Guarantees
OpenAI defended its position in a public blog post, contending that its agreement authorizes AI use for "lawful purposes" only. The company stated that it received assurance from the DoD that mass domestic surveillance is considered illegal and is explicitly excluded from the contract's scope.
Critics, however, warn that relying on current legality is risky, as laws regarding surveillance and warfare are subject to change. This legal ambiguity appears to be a central factor in Anthropic’s decision to decline the partnership.
Public Sentiment and User Reaction
The dispute appears to be influencing public perception. Following the announcement of OpenAI’s defense contract, data indicated a 295% spike in ChatGPT uninstallations. Amodei noted in his memo that the public largely views OpenAI’s alliance with the military with suspicion, while Anthropic has seen a corresponding rise in app store popularity. He expressed concern that OpenAI’s narrative might mislead their own employees, despite the negative media coverage and user pushback.
Comments
Leave a comment