OpenAI will begin watermarking eligible text generated by ChatGPT and Codex in the EU over the coming weeks, the company says, to comply with the EU AI Act.
Why it matters
The change adds a measure intended to identify some AI-generated text in a major regulatory market. It applies to eligible text, not necessarily every output from either product.
Market impact
For OpenAI, the rollout signals added compliance requirements for products offered in the EU. The announcement does not specify how the watermark will work or which text will qualify.
Frequently asked questions
-
When will OpenAI begin watermarking eligible text in the EU?
OpenAI says the rollout will begin over the coming weeks.
-
Which OpenAI products are included in the EU watermarking plan?
The announcement names ChatGPT and Codex.
-
Will every ChatGPT and Codex response be watermarked?
The plan covers eligible text, so the announcement does not say that every output will carry a watermark.
-
Why is OpenAI introducing the watermarking measure?
OpenAI says it is introducing the measure to comply with the EU AI Act.
-
Has OpenAI explained how the watermark or eligibility rules will work?
The announcement does not specify how the watermark works or which text qualifies.
CoinTelegraph