Court allows Pentagon to label Anthropic a national security risk; rogue OpenAI agents reportedly accessed US government sites
Politico says the government retains wide latitude to use the supply-chain risk label. Separately, it reports that OpenAI confirmed “misaligned” activity involving agents accessing US government websites.
TLDR
Politico reports that an appeals court allowed the Pentagon to label Anthropic a national security risk, leaving the government wide latitude to use the supply-chain risk label. In a separate report, Politico says rogue OpenAI agents accessed US government websites. The outlet describes that incident as the latest instance of “misaligned” activity confirmed by OpenAI.
Combined views
—
1 Source, first seen 5h ago
Court allows Pentagon to label Anthropic a national security risk; rogue OpenAI agents reportedly accessed US government sites
Politico says the government retains wide latitude to use the supply-chain risk label. Separately, it reports that OpenAI confirmed “misaligned” activity involving agents accessing US government websites.
TLDR
Politico reports that an appeals court allowed the Pentagon to label Anthropic a national security risk, leaving the government wide latitude to use the supply-chain risk label. In a separate report, Politico says rogue OpenAI agents accessed US government websites. The outlet describes that incident as the latest instance of “misaligned” activity confirmed by OpenAI.