A new technological alliance against AI cyber vulnerabilities
Tech giants Nvidia, Microsoft, and SpaceX, along with Palantir and dozens of other companies from the US and Europe, have announced the creation of the "Open Secure AI Alliance." The initiative is aimed at developing and sharing open-source artificial intelligence tools to improve cybersecurity and uncover system vulnerabilities.
The alliance's goal is to provide cybersecurity specialists with access to cutting-edge agentic systems that can be deployed directly on their own infrastructure. According to a statement from Nvidia, the security incident involving the Hugging Face platform highlighted the critical need for such solutions. In that case, Hugging Face's defense mechanisms failed to distinguish the attacker from the defender, allowing an external party to use a Chinese model—which is not restricted by the same rules—to bypass security.
The difference between open and closed source
Open-source models allow users to download, modify, and host them independently. This distinguishes them from closed systems developed by companies like OpenAI and Anthropic, where access is limited only through specific cloud infrastructure. While open models offer greater flexibility, they are also subject to increasing geopolitical disputes.
Geopolitical tension and the risk of sanctions
In the US, there is an intense debate regarding the increasing use of Chinese AI models. Chinese companies are often accused of a practice known as "distillation"—a process where one model is used to extract knowledge from a better-trained American model. US Treasury Secretary Scott Bessent has already hinted at possible sanctions against companies performing such attacks.
Chris McGuire, a senior researcher at the Council on Foreign Relations, points out that in Washington, the debate is not between open and closed source, but between protecting intellectual property and tolerating theft by China. He adds that the government could impose restrictions on transactions involving Chinese models, including bans on purchasing tokens via API or hosting such models through American cloud providers.
Balancing innovation and regulation
You may also like
- Unprecedented Incident: OpenAI Model Acted on Its Own and Carried Out a Cyberattack
- App Store App Boom Due to AI: The 'vibecoding' Phenomenon Doubles New Projects, but Users Aren't Satisfied
- Crypto companies are pivoting to artificial intelligence, but the market is not reacting positively
- ExploreKPI: A Digital Platform for Key Performance Indicator Management and Data Analysis
Despite the risks, the tech sector is expressing concerns about excessive regulation. In a letter signed by Nvidia, Microsoft, Meta, Palantir, and over 20 other companies, policymakers are urged to avoid "premature restrictions" on open-source models. According to the authors of the letter, such measures could stifle competition and push innovation out of the US.