Weasel Words: OpenAI's Pentagon Deal Won't Stop AI‑Powered Surveillance (www.eff.org)

OpenAI has come under fire for its recent contract with the U.S. Department of Defense (DoD), aimed at filling the gap left by rival Anthropic's refusal to support AI for surveillance. Following significant backlash from users and employees, which led to a nearly 300% increase in ChatGPT uninstalls, CEO Sam Altman acknowledged the deal was "opportunistic and sloppy." Although OpenAI has amended the contract to stipulate that its AI tools will not be used for domestic surveillance of U.S. nationals, this assurance is viewed skeptically due to the government's history of interpreting laws loosely to justify mass surveillance activities. The implications for the AI/ML community are considerable, as the language in OpenAI's agreements—filled with terms like "incidentally" and "unconstrained"—raises concerns about accountability and transparency. Critics argue that such vague wording allows the government to exploit legal loopholes for extensive surveillance under the guise of compliance. While OpenAI aims to steer government usage of AI toward democratic norms, the effectiveness of these measures is questionable given the historical context. The controversy highlights the ongoing tension between technological advancement and civil liberties, reinforcing the need for stronger legal protections and ethical considerations in AI deployment.