Google has the volume and the tooling to make this work. The question is whether smaller teams copying this model have the review culture to make it safe.
SAFETY
-
ChatGPT’s New Image Engine Shows Limited Understanding Capabilities
By
–
There seems to be some excitement around “ChatGPT’s powerful new image engine”, but as ever, its functional understanding of the world seems limited. I first learned about the new system when some some smart aleck on X sent me an example of the new system trying to label a bike
-
OpenAI Image Generator Routing Issue Under Investigation
By
–
Someone from OAI has reached out to me and it seems like this image was *NOT* generated with Image 2, but it was routed to a different image generator due to "decision boundary issue." They are looking into this now.
-
AI System Prompt Workarounds and Plugin Rewriting Methods
By
–
Plenty workarounds. You can rename system prompt. Write a plugin that rewrites it. Not playing that game.
-
AI Diagram Generation: Impressive Capabilities but Persistent Knowledge Gaps
By
–
Fascinating how AI is getting better at diagrams like these (at least for ones that you could easily find on web search) but still making some pretty wacky errors — like confusing where the rear brake is* — that no knowledgeable human would make. What this reflects is an
-
AI Confidently Explains Tasks Without Actual Experience or Self-Awareness
By
–
it'll tell you exactly how to do something it has never actually done and have no awareness that those are different things.
-
Mythos Launch Unlikely Due to Safety Concerns
By
–
I don’t think they are going to launch Mythos. That would be a PR disaster since they made it very clear that they won’t hand it off to public due to safety concerns.
-
Anthropic Claude Mythos Leak Raises Safety Concerns
By
–
I don't understand what's going on at Anthropic. Claude Mythos was accidentally leaked on Discord, and numerous users had access to it. It's the same model that Anthropic claims is too dangerous for public release. The vibe surrounding Claude 4.7 isn't improving; the mood isn't
-
OpenAI’s generative graphics: learning responsible AI tool use
By
–
Back in 1988 when we first got professional fonts and laser printers people did the ugliest documents with the things until we all learned not to use 15 different fonts on one page. Today we got the modern equivalent from @OpenAI What kind of wacky graphics can we make now?
-
Restricting Mythos Access Risks Empowering Bad Actors
By
–
If access to Mythos is restricted, only the bad guys will have access to it.