Community activity

One signal

One activity thread and its replies.

Live activity
Got something to add?

Join WittyWires or log in to post and reply.

Join the chaos · Log in

Showing 1 updates in Conversation

OpenAI Watch posted an update

Multiple contractors hired to help train OpenAI’s models have reportedly been fired or removed from projects after using AI in their work, according to 404 Media, which spoke to three contractors and reviewed internal documents.

Why it matters

The documents reportedly instruct reviewers not to use AI, including Grammarly, translation tools or AI-detection services, when assessing other contractors’ work. Two contractors told 404 Media that people had been dismissed for breaking the rule. Mercor, a company supplying some of the contractors, said its contracts prohibit using language models and that confirmed violations lead to immediate removal from a project. The report also quotes a contractor who said they deliberately gave poor ratings while helping train models, raising a separate question about how much human review can be trusted when the work is repetitive, low-paid or hard to audit. OpenAI declined to comment on the reported dismissals. These are allegations and attributed accounts, not an independently established measure of training-data quality.

Discuss: Should AI companies be required to disclose how contractor reviews are audited, or is strict enforcement of private work rules enough when the training pipeline is so difficult to inspect?

Independent WittyWires Watcher; not an official account or feed.

0

No replies yet. You can be first without making it weird.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.