@SecureBio

SecureBio is a research non-profit that works to advance biotechnology safely and prevent catastrophic pandemics

securebio.org
Joined January 2025
  • Tweets232
  • Following1
  • Followers1.1K
85Media
Clarifying our current engagement with Anthropic, and how we approach independence and publication rights in our work with AI developers.

Response to Anthropic's Update on Independent Evaluators

Article · @SecureBio
11
4
79
8,633
A common argument for downplaying the threat of deliberate biological misuse is that “nature is the worst bioterrorist”. In our latest blog, we argue that nature is the baseline, not the ceiling. Optimization for harm could produce something worse than a natural pandemic. Read more: securebio.substack.com/p/nat…
A six-panel black-and-white linocut-style illustration showing virus particles in the top row and flowers in the bottom row. Illustrated by Becca Chandler Bostock, 2026

ALT A six-panel black-and-white linocut-style illustration showing virus particles in the top row and flowers in the bottom row. Illustrated by Becca Chandler Bostock, 2026

5
280
We’ve long kept our core computational pipeline public to support the wider biosurveillance field. But as we’ve scaled, the security and operational costs of this approach have grown. With this in mind, we’re shifting future bioinformatics development to private repositories. Our existing public repositories aren’t going anywhere, but will no longer be regularly updated. We’ll continue to provide high-level updates to partners who rely on our analyses and to engage with researchers and practitioners to build a more secure and sustainable biosurveillance system. Read our full blog post: securebio.org/blog/taking-ou…
1
5
275
Frontier AI now beats the best human virologists on biosecurity-relevant evals. Once an AI model bests every human expert we can test it against, a higher score stops telling us anything about real-world risk. Read our new blog post on how bio evals have to adapt: securebio.org/blog/evaluatin…
Diagram showing that as AI model capabilities move from beginner to expert to superhuman, evaluations for chess and cybersecurity can shift from human-built proxies (puzzles, coding benchmarks) to real endpoints (real games, real exploits), while biosecurity cannot safely make that shift and must instead rely on indirect signals: experimental prediction, design tasks, defensive acceleration, and adjacent domains.

ALT Diagram showing that as AI model capabilities move from beginner to expert to superhuman, evaluations for chess and cybersecurity can shift from human-built proxies (puzzles, coding benchmarks) to real endpoints (real games, real exploits), while biosecurity cannot safely make that shift and must instead rely on indirect signals: experimental prediction, design tasks, defensive acceleration, and adjacent domains.

3
12
598
Today, more than 100 leading AI experts endorsed a set of minimum requirements to take seriously AI companies' recent call to embed external evaluators. These evaluators need to be genuinely independent, transparent, and represent a range of expertise areas. They also need to be guaranteed employee-level access and to be protected from retaliation for findings that make companies look bad. We welcome model developers’ recent calls for independent oversight, but it’s what they do next that matters. The labs must be accountable for ensuring these requirements are met, so that the public can have faith in the process and the outcomes. Over the past week, the AI community has debated the appropriate role of external evaluation, including who should do it and on what terms. We may not agree on everything, but there is a lot of common ground. To make embedded evaluations credible, more than 100 experts with varying backgrounds and ideas about AI risk agree in today’s letter that frontier AI developers should: 1. Guarantee embedded evaluators full editorial independence and mitigate conflicts of interest 2. Rely on multiple evaluators with differing viewpoints and areas of expertise 3. Publicly document the terms under which evaluators operate, as well as facilitating permissive publication of methods and findings 4. Shield evaluators from retaliation 5. Grant access equivalent to that of highly privileged employees There is a thriving and growing ecosystem of independent AI evaluators who are advancing this science every day – but we need aligned standards, guaranteed protections, and independent funding. That’s why we created the AI Evaluator Forum. Today we are entering our next phase. We’re launching an open call for new members, collaborators, and independent funding sources to help evaluators meet this moment and demand accountability from developers. Join us in building the evaluator ecosystem. See the public letter here: aievaluatorforum.org/initiat… Learn more at aievaluatorforum.org/path-ah…

Minimum Conditions for Embedding Evaluators

An open letter calling on all frontier AI companies to embed independent evaluators — signed by 200+ researchers, practitioners, and organization leaders.

aievaluatorforum.org
CNBC
@CNBCSep 18
Anthropic and OpenAI need truly independent safety evaluators, experts say in public letter cnbc.com/2026/09/18/ai-safet…

Anthropic and OpenAI need truly independent safety evaluators, experts say in public letter

A coalition of over 100 AI experts are urging independence and transparency from Anthropic, OpenAI and other foundation model labs to conduct evaluations.

cnbc.com
18
61
28
225
93,750
Untargeted wastewater metagenomic sequencing screens hundreds of pathogens at once, but that breadth costs coverage, making variant tracking more difficult. However, by pooling CASPER reads nationally, we can recover real lineage and genotype trends. Read more: securebio.org/blog/tracking-…
1
152
SecureBio AI has conducted pre-release assessments and independent risk evaluations for multiple frontier model companies. We are hiring for engineers, researchers, and many other roles to expand this work. Reach out if you're interested in working on bio-focused third-party assessment!
Dario Amodei
@DarioAmodeiSep 12
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so. Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training. You can read the full post here: darioamodei.com/post/we-must…

Dario Amodei — We Must Pace the Frontier

darioamodei.com
1
3
21
1,795
Current open roles: securebio.org/careers/
1
84
We ran a brief pre-release assessment of OpenAI's GPT-6 Astra that included manual testing and key benchmarks evaluating misuse-relevant knowledge, agentic capabilities, and refusal behavior for hazardous biology queries. Read our summary here: securebio.substack.com/p/sec…
1
4
217
more replies
With production safeguards on, GPT-6 Astra refused or blocked 82.9% of hazardous BioTIER queries (vs. 66% for GPT-5.6 Sol) while still complying with 95.3% of benign ones, and refused the screening-evasion task 100% of the time.
1
79
Ten hours of manual testing yielded similar findings, showing generally robust refusal behavior but some gaps, and a willingness to help with biological tool use and engage in high-level discussions. Our full report is here: securebio.org/resources/gpt-…
61
SecureBio retweeted
Patricia Paskov
@prpaskovSep 12
A non-exhaustive list of more orgs doing frontier eval work (I make no claims in either direction re their engagements with labs): @ApolloResearch @farairesearch @SecureBio @Irregular @ActiveSiteBio @redwood_ai @SaferAI_org @ValsAI @EpochAIResearch @AVERIorg AI Evaluator Forum has released AEF-one, Minimum Operating Conditions for Independent Third Party AI Evaluations: aef.one/aef-one.pdf / aievaluatorforum.org/ Guidelight AI has rolled out assessment standards: guidelight.ai/ @Fathom_org is working on building the auditing ecosystem, including through the newly launched @pact_ai_org So many hands on deck! So many more needed!
3
8
2
57
2,557
We updated our flagship virology troubleshooting benchmark, the Virology Capabilities Test (VCT), to be more shortcut-resistant and to have a larger headroom. We confirmed that the majority of VCT questions are scientifically valid and that neither VCT nor VCT-v2 is saturated. (Read the full post at securebio.substack.com/p/int…)
Flowchart showing theeneration of the updated 279-question VCT-v2. Of the 322 VCT questions, 43 were removed, including 25 easy, non-discriminating questions; 163 were edited to improve their scientific accuracy, clarity, and/or resistance to heuristic shortcuts; and 116 remained unchanged.

ALT Flowchart showing theeneration of the updated 279-question VCT-v2. Of the 322 VCT questions, 43 were removed, including 25 easy, non-discriminating questions; 163 were edited to improve their scientific accuracy, clarity, and/or resistance to heuristic shortcuts; and 116 remained unchanged.

1
1
5
390
more replies
We found that 27% of the original VCT questions were shortcut-exploitable (answerable without the image or question text). In all, we edited 163 questions (50.6%) and removed 43 questions (including 25 non-discriminating ones) to produce the 279-question VCT-v2 benchmark.
Flowchart showing the generation of the updated 279-question VCT-v2. Of the 322 VCT questions, 43 were removed, including 25 easy, non-discriminating questions; 163 were edited to improve their scientific accuracy, clarity, and/or resistance to heuristic shortcuts; and 116 remained unchanged.

ALT Flowchart showing the generation of the updated 279-question VCT-v2. Of the 322 VCT questions, 43 were removed, including 25 easy, non-discriminating questions; 163 were edited to improve their scientific accuracy, clarity, and/or resistance to heuristic shortcuts; and 116 remained unchanged.

1
1
66
Model accuracy dropped only 1.0–7.1 points on VCT-v2, and the gap between strongest and weakest models held steady, so past VCT results remain largely valid. VCT-v2 also has more headroom before saturation: at least ~30 points vs. ~23 on the original VCT.
Heatmap showing per-question performances of models and human experts on VCT-v2. Each cell represents the mean question × model accuracy, with darker colors indicating higher accuracy. Mean model accuracy on all questions is displayed to the right of the heatmap. Yellow cells represent data missing due to model refusals or missing expert baseline entries. The theoretical best model is taken to be the top-performing model on a per-question basis.

ALT Heatmap showing per-question performances of models and human experts on VCT-v2. Each cell represents the mean question × model accuracy, with darker colors indicating higher accuracy. Mean model accuracy on all questions is displayed to the right of the heatmap. Yellow cells represent data missing due to model refusals or missing expert baseline entries. The theoretical best model is taken to be the top-performing model on a per-question basis.

1
55
Load more