Scale AI
@scale_AI
Followers
67K
Following
1K
Media
508
Statuses
2K
To make the best models, you need the best data.
Joined July 2016
RT @calvincbzhang: New @scale_AI research in collaboration with @AnthropicAI introduces SHADE-Arena, a benchmark to test for AI sabotage. S….
0
6
0
Developed with guidance from our expert red-teamers, FORTRESS sets a new standard for evaluating LLMs. Especially where the stakes are highest.
1
0
10
As AI capabilities grow, so do the risks. FORTRESS is the first benchmark designed to evaluate both sides of the dual-use national security spectrum:. ⚠️ Can models be misused for harm?.🛑 Are they over-restricting safe, critical information?.
1
0
5
Introducing FORTRESS. Our newest benchmark built to evaluate AI models where it matters most: national security and public safety.
4
12
60
RT @SeanHendryx: What will the learning environments of the future look like that train artificial super intelligence? In recent work at @s….
0
29
0
RT @summeryue0: 🔍 SEAL and Red Team at @scale_ai present a position paper outlining what we’ve learned from red teaming LLMs so far—what ma….
0
21
0
The latest episode of Human in the Loop from Scale unpacks how to red team for real-world enterprise risk: model drift, over-restrictive guardrails, and agentic AI failures.
1
1
17
The enterprise AI threat landscape is evolving rapidly, but with red-teaming, so can your defenses
4
5
26
DeepSeek-R1's latest upgrade is now live on SEAL Leaderboards — outperforming all other open-source models on Humanity's Last Exam (Text Only).
5
10
32