Pentagon integrates Grok chatbot into its networks despite scandals surrounding the service

According to the Associated Press, Grok is set to launch at the U.S. Department of Defense later this month. Hegset made the announcement during a speech at the SpaceX facility in South Texas and promised to make «appropriate» data sets from military IT systems available for AI use. This includes both unclassified and classified networks, as well as data from intelligence bases.

The announcement came amid criticism of Grok for generating highly sexualized deepfakes of people without their consent. The AP also mentions previous scandals related to the chatbot’s anti-Semitic responses. After the outcry, Malaysia and Indonesia blocked access to Grok, and the British online safety watchdog announced the launch of an investigation. The developers responded by limiting image generation and editing to paid accounts.

Hegset stressed that he wants to abandon models that «do not allow for war,» and said that military AI would operate without «ideological constraints.» The AP article notes that under the Biden administration, in late 2024, national security agencies were allowed to expand the use of the most powerful models, but at the same time prohibited certain scenarios, including those that could violate constitutional rights or automate the use of nuclear weapons; it is unclear whether these restrictions are in effect now.

Grok is integrated into the X social network and is being pitched by Musk as an alternative to Gemini and ChatGPT. The Pentagon is simultaneously trying to scale generative AI across different departments to work faster with documents, analytics, and data, but the tool’s integration with reputational risks is already raising serious questions about security and control.

Previously, dev.ua wrote about how, after a high-profile scandal, Elon Musk’s company xAI urgently restricted access to the image generation function in the Grok chatbot. From now on, only paid subscribers are allowed to create images.

Grok generates thousands of deepfakes every hour — study

On the topic

Grok generates thousands of deepfakes every hour — study

Elon Musk’s AI got into a sex scandal: Grok started «undressing» children in photos

On the topic

Elon Musk’s AI got into a sex scandal: Grok started «undressing» children in photos

AI Grok is ready to sacrifice half of humanity and all Jews to save Elon Musk

On the topic

AI Grok is ready to sacrifice half of humanity and all Jews to save Elon Musk

Read the country’s main IT news in our Telegram

On the topic

Read the country’s main IT news in our Telegram