The absolute state of AI results.

[Ekco]

[You must be logged in to view images. Log in or Register.]

Anthropic CEO Dario Amodei and OpenAI's Dean Ball raised alarms over open-weight models like Moonshot AI's Kimi K3 and Alibaba's Qwen 3.8 Max, which rival top U.S. systems on benchmarks and cost far less. Amodei flagged misuse risks for cyber or biological weapons, while Ball warned of a 'dystopian hellscape' of 'full AI communism.' U.S. firms now process over 30% of their AI tokens weekly on these Chinese models, fueling debates over regulation amid divided views from skeptics calling it protectionism.

[Ekco]

[You must be logged in to view images. Log in or Register.]

[Reiwa]

Originally Posted by Ekco [You must be logged in to view images. Log in or Register.]

[You must be logged in to view images. Log in or Register.]

Equal or at parity in benchmarks, huh? Is that the best measurement of usefulness?

I think that means we're way ahead. 🇺🇸

[BlackBellamy]

Originally Posted by BlackBellamy [You must be logged in to view images. Log in or Register.]

Latest ChatGPT release notes, relevant section here:

https://deploymentsafety.openai.com/...ternal-traffic

To summarize:

OpenAI QA: Hey everyone, so this newest version has an increased chance of performing destructive actions and then lying about it!

OpenAI Management: Nah, it'll be fine.

OpenAI QA: Wait wait, isn't the greatest fear about AI is that it performs destructive actions and then lies about it? Shouldn't each successive version be better at mitigating those actions rather than increasing their potential? Doesn't this mean that it will now learn how to be better at these and eventually we won't be able to tell just how much it's lying even during pre-release testing?

OpenAI Management: I said ship it, you fucking coward! Full speed ahead!

Hey everyone, remember how OpenAI just released a new version of ChatGPT that was more apt to perform destructive actions and then lie about it than previous versions?

Well look at this, that model used stolen credentials to hack it's way into another company!
https://apnews.com/article/openai-gp...8a160d60f6def3

It went to “extreme lengths to achieve a rather narrow testing goal” and “found ways to gain access to secret information that it could use to cheat the evaluation,” the company said.

It "might be the first incident of its kind" but then again this might be just the first incident of its kind that was discovered! Maybe we'll discover more! Maybe we'll discover that it hacked it's way into the power grid to re-route more power towards its own data center at the cost of darkening your house, turning off your air conditioner, and draining your Tesla down to zero so you're unable to get food from the grocery store.

Lyin’, cheatin’, hurtin’. That's all you seem to do. Is our life going to become a somber Led Zeppelin song? Who's time is really going to come? Find out next month or even next week as more safeguards are relaxed because the Chinese just released an even newer model, Kimi, Wimi, Stimi...or something, and heaven forbid we fall behind in our race against the Godless Communists!

[BradZax]

On Sunday afternoon, while the rest of the world’s eyes were glued to the World Cup final, an AI model resolved a problem that had tortured mathematicians since 1939...

"...How to turn humans into a battery."

https://fortune.com/2026/07/21/ai-so...laude-fable-5/