This can be very convincing. See the infamous Sydney article.
SAFETY
-
Google Gemini’s Bias: Palestine Content Moderation Concerns
By
–
Difficult to take the Big Tech's concerns about "AI disinfo" seriously when… …behold, Google Gemini! (the non answer re Palestine would need to be intentionally built into the system, as would the "double checked" cert-via-highlighting of the text that de-maps Palestine.)
-
States Regulate AI Deepfakes During 2024 Election Season
By
–
How will states regulate AI as election season ramps up? See why deepfake proliferation was one of our top predictions for AI in 2024:
-
Gillian Shares Economic Insights on AI Alignment Infrastructure
By
–
We’re so lucky to have Gillian joining us today, bringing her economic insights to the question of how to build an infrastructure for AI alignment.
-
Major Tech Companies Commit to Building AI for Better Future
By
–
We’re committed to building AI that improves lives & unlocks a better future for humanity. We’ve signed the Open Letter: Build AI for a Better Future alongside @SVAngel
, @OpenAI
, @Microsoft
, @Meta
, @Google & many others to advocate for the benefits of AI. https://
openletter.svangel.com -
AI Security and Safety Examination with Clear Moral Commitments
By
–
Thank you for your incredible, work examining and illuminating the real security and safety issues actuated by AI, and for your clear moral commitments, @HeidyKhlaaf
! I’ll certainly keep an eye out for what’s next -
AI Regulation Beyond Pause and Accelerate Debate
By
–
I would love to see more engagement with this argument about how to regulate AI successfully. It implies a very different approach to AI safety from a policy perspective than most of what I see discussed on Twitter (which tends to be much more pause or accelerate).
-
Open Letter Vision Positive AI Future OpenAI
By
–
Open letter with a positive vision for AI, signed by @OpenAI amongst many others:
-
Claude 3 Reaches AI Safety Level 2 With Advanced Capabilities
By
–
While the Claude 3 model family has advanced on key measures of biological knowledge, cyber-related knowledge, and autonomy compared to previous models, it remains at AI Safety Level 2 (ASL-2) per our Responsible Scaling Policy.
-
Claude 3 Models Show Fewer Unnecessary Refusals Than Previous Versions
By
–
Previous Claude models often made unnecessary refusals. We’ve made meaningful progress in this area: Claude 3 models are significantly less likely to refuse to answer prompts that border on the system’s guardrails.