The US military has ramped up its use of artificial intelligence tools after the October 7 Hamas attacks on Israel, based on a new report by Bloomberg. Schuyler Moore, US Central Command's chief technology officer, told the news organization that machine learning algorithms helped the Pentagon identify targets for more than 85 air strikes in the Middle East this month. 

US bombers and fighter aircraft carried out those air strikes against seven facilities in Iraq and Syria on February 2, fully destroying or at least damaging rockets, missiles, drone storage facilities and militia operations centers. The Pentagon had also used AI systems to find rocket launchers in Yemen and surface combatants in the Red Sea, which it had then destroyed through multiple air strikes in the same month.

The machine learning algorithms used to narrow down targets were developed under Project Maven, Google's now-defunct partnership the Pentagon. To be precise, the project entailed the use of Google's artificial intelligence technology by the US military to analyze drone footage and flag images for further human review. It caused an uproar among Google employees: Thousands had petitioned the company to end its partnership with Pentagon, and some even quit over its involvement altogether. A few months after that employee protest, Google decided not to renew its contract, which had ended in 2019. 

Moore told Bloomberg that US forces in the Middle East haven't stopped experimenting with the use of algorithms to identify potential targets using drone or satellite imagery even after Google ended its involvement. The military has been testing out their use over the past year in digital exercises, she said, but it started using targeting algorithms in actual operations after the October 7 Hamas attacks. She clarified, however, that human workers constantly checked and verified the AI systems' target recommendations. Human personnel were also the ones who proposed how to stage the attacks and which weapons to use. "There is never an algorithm that’s just running, coming to a conclusion and then pushing onto the next step," she said. "Every step that involves AI has a human checking in at the end."

The US military has ramped up its use of artificial intelligence tools after the October 7 Hamas attacks on Israel, based on a new report by Bloomberg. Schuyler Moore, US Central Command's chief technology officer, told the news organization that machine learning algorithms helped the Pentagon identify targets for more than 85 air strikes in the Middle East this month. 

US bombers and fighter aircraft carried out those air strikes against seven facilities in Iraq and Syria on February 2, fully destroying or at least damaging rockets, missiles, drone storage facilities and militia operations centers. The Pentagon had also used AI systems to find rocket launchers in Yemen and surface combatants in the Red Sea, which it had then destroyed through multiple air strikes in the same month.

The machine learning algorithms used to narrow down targets were developed under Project Maven, Google's now-defunct partnership the Pentagon. To be precise, the project entailed the use of Google's artificial intelligence technology by the US military to analyze drone footage and flag images for further human review. It caused an uproar among Google employees: Thousands had petitioned the company to end its partnership with Pentagon, and some even quit over its involvement altogether. A few months after that employee protest, Google decided not to renew its contract, which had ended in 2019. 

Moore told Bloomberg that US forces in the Middle East haven't stopped experimenting with the use of algorithms to identify potential targets using drone or satellite imagery even after Google ended its involvement. The military has been testing out their use over the past year in digital exercises, she said, but it started using targeting algorithms in actual operations after the October 7 Hamas attacks. She clarified, however, that human workers constantly checked and verified the AI systems' target recommendations. Human personnel were also the ones who proposed how to stage the attacks and which weapons to use. "There is never an algorithm that’s just running, coming to a conclusion and then pushing onto the next step," she said. "Every step that involves AI has a human checking in at the end."

Just a few days ago, OpenAI's usage policies page explicitly states that the company prohibits the use of its technology for "military and warfare" purposes. That line has since been deleted. As first noticed by The Intercept, the company updated the page on January 10 "to be clearer and provide more service-specific guidance," as the changelog states. It still prohibits the use of its large language models (LLMs) for anything that can cause harm, and it warns people against using its services to "develop or use weapons." However, the company has removed language pertaining to "military and warfare."

While we've yet to see its real-life implications, this change in wording comes just as military agencies around the world are showing an interest in using AI. "Given the use of AI systems in the targeting of civilians in Gaza, it’s a notable moment to make the decision to remove the words ‘military and warfare’ from OpenAI’s permissible use policy,” Sarah Myers West, a managing director of the AI Now Institute, told the publication. 

The explicit mention of "military and warfare" in the list of prohibited uses indicated that OpenAI couldn't work with government agencies like the Department of Defense, which typically offers lucrative deals to contractors. At the moment, the company doesn't have a product that could directly kill or cause physical harm to anybody. But as The Intercept said, its technology could be used for tasks like writing code and processing procurement orders for things that could be used to kill people. 

When asked about the change in its policy wording, OpenAI spokesperson Niko Felix told the publication that the company "aimed to create a set of universal principles that are both easy to remember and apply, especially as our tools are now globally used by everyday users who can now also build GPTs." Felix explained that "a principle like ‘Don’t harm others’ is broad yet easily grasped and relevant in numerous contexts," adding that OpenAI "specifically cited weapons and injury to others as clear examples." However, the spokesperson reportedly declined to clarify whether prohibiting the use of its technology to "harm" others included all types of military use outside of weapons development. 

In a statement to Engadget, an OpenAI spokesperson admitted that the company is already working with the US Department of Defense. "Our policy does not allow our tools to be used to harm people, develop weapons, for communications surveillance, or to injure others or destroy property," the spokesperson said. "There are, however, national security use cases that align with our mission. For example, we are already working with DARPA to spur the creation of new cybersecurity tools to secure open source software that critical infrastructure and industry depend on. It was not clear whether these beneficial use cases would have been allowed under 'military' in our previous policies. So the goal with our policy update is to provide clarity and the ability to have these discussions."

Update, January 14 2024, 10:22AM ET: This story has been updated to include a statement from OpenAI.

Just a few days ago, OpenAI's usage policies page explicitly states that the company prohibits the use of its technology for "military and warfare" purposes. That line has since been deleted. As first noticed by The Intercept, the company updated the page on January 10 "to be clearer and provide more service-specific guidance," as the changelog states. It still prohibits the use of its large language models (LLMs) for anything that can cause harm, and it warns people against using its services to "develop or use weapons." However, the company has removed language pertaining to "military and warfare."

While we've yet to see its real-life implications, this change in wording comes just as military agencies around the world are showing an interest in using AI. "Given the use of AI systems in the targeting of civilians in Gaza, it’s a notable moment to make the decision to remove the words ‘military and warfare’ from OpenAI’s permissible use policy,” Sarah Myers West, a managing director of the AI Now Institute, told the publication. 

The explicit mention of "military and warfare" in the list of prohibited uses indicated that OpenAI couldn't work with government agencies like the Department of Defense, which typically offers lucrative deals to contractors. At the moment, the company doesn't have a product that could directly kill or cause physical harm to anybody. But as The Intercept said, its technology could be used for tasks like writing code and processing procurement orders for things that could be used to kill people. 

When asked about the change in its policy wording, OpenAI spokesperson Niko Felix told the publication that the company "aimed to create a set of universal principles that are both easy to remember and apply, especially as our tools are now globally used by everyday users who can now also build GPTs." Felix explained that "a principle like ‘Don’t harm others’ is broad yet easily grasped and relevant in numerous contexts," adding that OpenAI "specifically cited weapons and injury to others as clear examples." However, the spokesperson reportedly declined to clarify whether prohibiting the use of its technology to "harm" others included all types of military use outside of weapons development. 

In a statement to Engadget, an OpenAI spokesperson said, "Our policy does not allow our tools to be used to harm people, develop weapons, for communications surveillance, or to injure others or destroy property. There are, however, national security use cases that align with our mission. For example, we are already working with DARPA to spur the creation of new cybersecurity tools to secure open source software that critical infrastructure and industry depend on. It was not clear whether these beneficial use cases would have been allowed under “military” in our previous policies. So the goal with our policy update is to provide clarity and the ability to have these discussions."

Update, January 14 2024, 10:22AM ET: This story has been updated to include a statement from OpenAI.