Cybernews researchers say that out of 10 AI models using public information, five have B or lower risk. The remaining five, including anthropomorphism, cohere and mistral, were rated as low risk.
Two major participants, Openai and 01.AI, received D scores, indicating high risk, while the inflection point AI scored F, which is a serious security risk.
Additionally, five of the 10 providers recorded data breaches, the researchers said. They said Openai was reportedly suffering the biggest violation, with 1,140 incidents, including the recent data breach only nine days before the analysis was conducted. Researchers say AI confusion AI experienced violations 13 days ago.
