Security researchers have discovered a vulnerability in the APIs of major AI providers OpenAI, Anthropic, and Google that allows extraction of encrypted reasoning traces and their transfer between different AI models, according to The Decoder. This flaw reveals that the reasoning summaries presented to users often mask the actual processes performed by the models.
The Decoder also reported that a scan of public sessions uncovered dozens of leaked passwords and API keys, raising concerns about potential security breaches through these AI platforms. The exposed data could pose risks for users relying on these services for sensitive or proprietary operations.
For Japanese markets, where AI integration in FX trading and crypto analytics is growing, this vulnerability highlights the critical need for robust security measures around AI tools to protect confidential data and maintain trust in automated decision-making systems.