Peter Thiel, the Palantir co-founder and prominent Trump backer, claimed at the Aspen Ideas Festival on June 30, 2026, that Anthropic, which he called a 'woke liberal company,' would use its AI models to 'rig the elections in 2028,' a claim CNN's on-the-record report of the panel called 'unsupported' and 'conspiratorial.' Anthropic declined to comment, pointing instead to its April 2026 blog post on election safeguards, which details bias-testing and abuse-detection systems built into Claude. Thiel made the remarks during a wide-ranging discussion with political scientist Francis Fukuyama that also touched on real prior friction between Anthropic and the administration: the Pentagon designated Anthropic a 'supply chain risk' in March 2026, prompting a lawsuit, and CEO Dario Amodei has previously likened Trump to a 'feudal warlord.' For AI practitioners, the episode underscores that unverified political claims about model misuse still carry real regulatory and reputational stakes.
Trump unveils voluntary AI safety accord with six tech giants