OpenAI plans invisible watermark for eligible ChatGPT text in the EU
Over the coming weeks, with no date given, OpenAI will add an invisible watermark to eligible ChatGPT and Codex text generated in the EU, under the AI Act.
Source of truthcommunity.openai.com/…/1403521On October 5 OpenAI explained how it plans to implement the AI Act's text provenance rules. From that day, API customers can opt in to text watermarking for select models. Over the coming weeks, with no date given, the company will add an invisible watermark to ‘eligible’ ChatGPT and Codex text generated in the EU.
The technology is called textGrain and adds a statistical signal through the words the model chooses. According to OpenAI, it can indicate that one of its systems generated or processed part of a text, but not who used it or whether the content is accurate. The watermark is invisible, and detector access will initially be limited to approved researchers and expert organizations.
Related stories
Microsoft: Storm-0940 group automated attacks with AI agents
According to Microsoft, Storm-0940, which it assesses as China-state aligned, used the open-source ARTEX framework to automate attacks targeting hundreds of organizations; most of the targeting was not successful.
LiteLLM: malicious version 1.107.0 published to PyPI and Docker
According to LiteLLM, version 1.107.0, pushed on October 21, contained malware that steals credentials. LiteLLM advises anyone who installed it on a host with outbound internet access to assume the data was exfiltrated and to rotate keys and secrets.