Experts from Endor Labs are weighing in on the debate surrounding AI transparency versus so-called open-washing. Andrew Stiefel, Senior Product Marketing Manager at Endor Labs, stressed the importance of applying lessons from software security to AI.
He referenced the U.S. government’s 2021 Executive Order on Improving America’s Cybersecurity, which requires a software bill of materials (SBOM) for products sold to federal agencies.
An SBOM is an inventory that lists all open-source components, helping to detect vulnerabilities. Stiefel argued that using similar practices for AI systems would enhance transparency by revealing a model’s datasets, training processes, weights, and other core elements, thus improving security for both citizens and government workers.
Julien Sobrier, Senior Product Manager at Endor Labs, added that the definition of an open AI model is complex. He explained that a truly open model should include every element, from the training set and weights.
He noted that industry giants like OpenAI and Meta have stirred confusion by offering models that are only partially open. Sobrier warned against open-washing, where companies claim transparency while imposing restrictions to protect competitive interests, citing examples of cloud providers offering paid versions of open-source projects with commercial limitations.
The discussion also highlighted DeepSeek, a rising player in the AI field, which has released its models and weights as open-source. This move, according to Stiefel, allows for better auditing of security risks and offers a blueprint for safe AI deployment.
As more organizations adopt open-source AI, developing common standards to evaluate security, quality, and operational risks becomes essential. Both experts agreed that balancing innovation with strict risk management is key to fostering a responsible AI future, ensuring that progress in the field does not come at the expense of security or ethical standards.
Stiefel and Sobrier agree that clear guidelines and best practices are vital for ensuring innovation, transparency, and safety.
