News for hackers. Stuff that matters.

A comment on DeepSeek R1 Appears to Be a CCP Cyberespionage Stunt Involving Possible Murder.

Re:DeepSeek R1 Appears to Be a CCP Cyberespionage Stunt Invo... by mrkeen on Wednesday January 29, 2025 @06:20AM (#42863737)
Irrelevant since it's all closed anyway.

Who cares if they claimed they trained it using only a paper cup and two paper-clips. Their model works exactly as well as it works, and you're not reproducing their work either way.

[ Parent | on HN ]
Re:DeepSeek R1 Appears to Be a CCP Cyberespionage Stunt Invo... by Jasondells on Wednesday January 29, 2025 @06:30AM (#42863787)
The model works, sure, but that's not the point. OpenAI and others can replicate any technical advances for future models moving forward. DeepSeek R1 will be "outdated" in 6-12 months and be replaced by other, more powerful models for sure. If DeepSeek's success is based on stolen data, they won't be able to replicate their success as for sure such security gaps will now get closed.

There have been talks about a laissez-faire attitude regarding cybersecurity at OpenAI for a long time.... but this is surely coming to an end now. Same at Google.

[ Parent | on HN ]