EU presses OpenAI, Anthropic for more details on AI safety incidents

AI giants must hand over information about their models under EU rules

The European Commission has requested additional information from American tech giants OpenAI and Anthropic regarding recent safety incidents, a spokesperson said on Monday.

In recent months, both companies have been involved in AI safety incidents that made headlines across the world.

"The questions focus on how many incidents happened, where the EU was affected and how providers have changed their risk assessment, security and safety practices in response to these incidents," a Commission spokesperson told Rapporteur.

Additionally, the EU's executive wants information about reports that new frontier models are becoming technically harder to monitor, they added.

Under the EU's AI Act, AI labs must take appropriate measures to make their technologies safe. The Commission can demand information to check they're sticking to the rules.

If serious safety incidents happen, model developers must also inform the Commission about them.

Back in September, a Commission spokesperson warned AI labs that the legal requirement was not just a "tick box" exercise, saying that responses must be "precise and accurate".