@LRudL_: I think it's possible to make a lot of progress toward models that are both open and safe! The former will become more …
Summary
A tweet discussing the potential to advance AI models that are both open and safe, highlighting the importance of open-source to prevent centralization and safety due to current challenges.
View Cached Full Text
Cached at: 09/02/26, 09:54 AM
I think it’s possible to make a lot of progress toward models that are both open and safe!
The former will become more important to avoid centralization & spread benefits, and the second will become more important as gestures at state of things
Alex Robey (@AlexRobey23): We’re hiring safety researchers at @thinkymachines! 🦺
My team works on safety across the model development stack – pre-training data filtering, harmful capability evals, safety post-training, red-teaming, abliteration/malicious fine-tuning.
We’re particularly interested in
Similar Articles
Should public be barred from accessing extremely powerful models for fear of bad actors? Is open source reckless?
The article discusses the dilemma of whether to restrict access to powerful AI models to prevent misuse by bad actors or to open-source them for equitable access, weighing the risks of power consolidation vs. societal harm. It suggests a middle ground, citing Anthropic's approach with guardrails, but acknowledges the limitations and trade-offs.
@levie: In a world where there are strong open source alternatives that are only just behind the frontier models, you make your…
A commentary arguing that restricting access to frontier AI models in favor of closed-source alternatives is counterproductive, as strong open-source models close the gap and foreign ecosystems benefit from openness.
@johnschulman2: We love open weights and plan to keep releasing open-weight models and fine-tuning tools. But we’re not absolutists; mi…
John Schulman discusses a balanced approach to open-weight AI models, acknowledging misuse risks while committing to continued releases and calling for collaborative research on safe access.
@MTSlive: We asked the CEO of HuggingFace @ClementDelangue what the risks of releasing powerful open source models are. He says r…
HuggingFace CEO Clément Delangue argues that restricting open source AI models creates more risk than openness, citing historical examples like GPT-2 and Mythos to support his view that openness improves cybersecurity and overall safety.
@TheAhmadOsman: That's why:
A tweet by @TheAhmadOsman advocates for open-source AI, arguing that artificial intelligence must remain accessible and community-governed to avoid dependency on closed corporate systems.