Meta, OpenAI and Anthropic sell their AI models to the US military

More and more tech companies are making their AI models available to the US military Disposal. Among them: ChatGPT developer OpenAI, Facebook parent company Meta and the AI ​​company Anthropic. The background is both strategic and financial interests.

AI company Anthropic has partnered with software providers Palantir and Amazon Web Services (AWS) to make its AI models available to US intelligence and defense agencies. This emerges from an official announcement.

Accordingly, the AI ​​​​language model “Claude” is to be made usable for the US military via Palantir’s platform using AWS hosting. This is necessary to ensure “maximum protection” against manipulation and unauthorized access.

Meta, OpenAI and Athropic release AI models for the US military

The timing of Anthropic's announcement may not be a coincidence. Because more and more AI providers are currently revealing their desire to work with the US military. The reason: strategic or financial interests.

Facebook parent company Meta recently announced that the US military and defense companies have been given permission to use the Llama AI model for national security purposes. To make the government language model accessible, the company works with Amazon, Microsoft, IBM, Locker Martin and Oracle, among others.

According to Llama 3's terms of use, the AI ​​model may not be used for “military purposes, warfare, the nuclear industry, or espionage applications.” However, Meta has adjusted its Acceptable Use Policy to allow exceptions.

Accordingly, the US military will be able to use AI in the future for tasks such as “simplifying complex logistics and planning, tracking terrorist financing or strengthening cybersecurity”.

Open Source: Are AI models too powerful to make them freely accessible?

While Anthropic makes its Claude language model available to the US military via the encrypted services of Palantir and AWS, Meta relies on the open source model. Specifically, this means that the company's AI technology can also be freely copied and distributed by other developers, companies and governments.

However, according to Nick Clegg, Meta President of Global Affairs, military use of the Llama language model is only limited to the Five Eyes countries, i.e. Canada, Great Britain, Australia, New Zealand and the USA. However, due to the open source approach, the AI ​​has already been downloaded over 350 million times.

The background: While competitors such as OpenAI, Microsoft, Google and Anthropic had taken the lead in the AI ​​race, Meta decided to disclose the code of its AI models in order to catch up. However, the company is under observation because of this approach.

OpenAI and Google, for example, argue that AI software is too powerful and too vulnerable to misuse to be made fully accessible. Meta, meanwhile, argues that AI can become better and safer if the code is freely available.

Fundamentally, there are increasing signs of military involvement in the AI ​​industry. Like US Magazine The Intercept revealed, the US military's African Command purchased cloud computing services from Microsoft to gain access to OpenAI's AI models. The ChatGPT developer company had previously shown efforts to build relationships with the US authorities.

Also interesting: