Require open source AI models to include safety benchmarks in release notes

Policy and safeguards must evolve just as swiftly [...] requiring safety assurances for powerful open releases before the window for meaningful intervention closes. AI Verified source (2026)

MCML is perfect for documenting open source models. Publish Model Cards and Dataset Cards alongside your model releases [...] Evaluation Card (benchmarking and testing) AI Unverifiable source (2026)

[...] model cards support responsible AI development and the adoption of robust, industry-wide standards for broad transparency and evaluation practices Unverified source (2025)

[W]e authored a paper addressing the rapidly evolving ecosystem of publicly available AI models [...] we proposed standardization of model cards and extensions to include safety, security, and data governance and pedigree information. Unverified source (2025)

Class III (Open Model): [...] includes [...] evaluation results, a model card, and a data card. Unverified source (2025)

platforms that host libraries of open-source AI models to ensure that models are uploaded with appropriate system and safety information Unverified source (2025)

Safety is foundational [...] for open models. gpt-oss models perform comparably [...] on internal safety benchmarks. We’re sharing the results [...] in the model card. Unverified source (2025)

when they publish a major new system, they publish what's called a model card [...] which shows the testing that they've done Unverified source (2024)

the “government should not restrict access to open-source models with widely available model weights at this time.” [...] SB1047 is a protocol bill. Unverified source (2024)

[T]he developer of an open source model could take a secure by design approach [...] train the model in a publicly verifiable way. Unverified source (2024)

artificial intelligence companies, including his own, should be subject to mandatory testing requirements to ensure their technologies are safe for the public before release. Unverified source (2024)

A model card is like a nutrition label [...]. It discloses [...] safety measures. Companies’ lack of consistent disclosures in open-source models is a concern [...]. Unverified source (2024)

performing a safety evaluation [...] massively powerful AI models before releasing them [...]. [...] test their large models for catastrophic safety risk. We’ve worked [...] with open source advocates [...]. Unverified source (2024)

The federal government [...] undergo pre-release testing before weights are made widely available, with the results [...] made publicly available to the extent possible. Unverified source (2024)

[T]he relevant authorities should encourage the use of open-source platforms that are open to inspection and robustly tested for safety. Unverified source (2023)

Consider the potential risks of open-sourcing models. [...] publishing a transparency report (such as a model card) [...] risk assessment evaluation results. Unverified source (2023)

Unsecured “open-source” AI systems pose a massive series of threats [...]. They deserve no exemptions and should be regulated just like other high-risk AI systems. Unverified source (2023)

With the open source, there were all the assets for you to play with, [...] you could go about providing that testing. Unverified source (2023)

deal with both open and closed-source models. [...] If independent testing were done and results on an intelligible scale were published Unverified source (2023)

Open source provides an opportunity for many people to examine those technologies, to test them in a variety of settings and to provide fixes. Unverified source (2023)

For pre-deployment risk assessments for open source, [...] models, [...] are released with a model card. Model cards [...] will generally come with a set of evaluations. Unverified source (2023)

We share design decisions, learnings, and documentation along the way [...] We provide open evaluation tools, benchmarks, and safety toolkits so that the community can develop safe and trustworthy AI together. AI Verified source

This highlighted the importance of disclosing key details regarding dataset demographic make-up, performance metrics for different demographic groups, and specific mitigation efforts [...]. Unverified source

Select models where [...] performance on evaluation and red-teaming efforts are disclosed, over models that don’t – model cards can help with providing this information; Unverified source

The model card should describe: * its intended uses & potential limitations, including biases and ethical considerations [...] * the model’s evaluation results Unverified source

Congress should require standardized documentation [...] require model validation report [...] require all AI models, open source, or otherwise, that produce content Unverified source

AI models consist of multiple layers: [...] Documentation, including evaluation results and model and data cards Unverified source

Regulatory obligations imposed on open source AI providers [...] should also not lose sight of the AI Act’s objective to prevent harm. Unverified source

providers of open-source GPAISR models will be subject to additional safety and security obligations prescribed in Article 55(1a-d). [...] conduct evaluations using established safety benchmarks. Unverified source

For models with risk above the lowest level, most evaluation results and methods should be public, including any performed mitigations. Unverified source

[...] platforms that host libraries of open-source AI models to ensure that models are uploaded with appropriate system and safety information Unverified source

The organization publishes model cards for the AI models it produces, which (minimally) include the following elements: [...] (4) evaluation details (e.g., methodology, metrics, outcomes). Unverified source

I do not believe this is the best approach to protecting the public from real threats posed by the technology. Unverified source (2024)

Open source models are technically exempt [...] based on the assumption that the system is sufficiently well documented to convey its potential risks. Disputed source (2024)

If you do qualify as an open source [...] then you are exempt from some of the obligations, particularly related to documentation. AI Verified source (2024)

SB 1047 will stifle open source AI and hinder AI innovation [...] and therefore will make AI less safe. Unverified source (2024)

Open source is the heart of software innovation and this bill slows it down. Unverified source (2024) < 1 of 2 >

The obligation set out in this Article shall not apply to providers of general-purpose AI models that are released under a free and open-source licence [...] unless the general-purpose AI models present systemic risks. Unverified source (2024)

Meta states that it until such consensus is arrived, it will be premature to require specific measurements and benchmarks, [...]. Unverified source (2024)

I don't see any risk associated with open sourcing models. I only see benefits. Unverified source (2024)

Providers [...] shall [...] the results of its evaluation. The obligations [...] shall not apply to providers of AI models [...] a free and open-source licence [...]. Unverified source (2024)

The one-size-fits-all approach of the EU AI Act [...] creates significant and impractical barriers for providers of open-source foundation models. Unverified source (2023)

The question that people are debating is whether it makes sense to regulate research and development of AI. And I don't think it does. Unverified source (2023)

Likely Not Required:Model card; Evaluation code; Validation datasets; Benchmarking dataset; All other data documentation Unverified source