Artificial intelligence used extensively in Gaza genocide: Mistook students for 'potential terrorists'

An investigation by the Associated Press (AP) has revealed that the Israeli army is making extensive use of artificial intelligence technologies in its operations in Gaza. Microsoft's Azure cloud platform and OpenAI models have played a significant role in the Israeli army's data analysis and target identification processes. According to the investigation, the Israeli army increased its use of artificial intelligence by 200 times following the attacks on October 7, 2023. As of March 2024, the amount of data stored on Microsoft servers exceeded 13.6 petabytes.

Documents reviewed by the AP indicate that a three-year contract worth 133 million dollars was signed between Microsoft and the Israeli Ministry of Defense in 2021. This agreement made Israel Microsoft's second-largest military customer after the United States.

An Israeli intelligence official stated that the army analyzes data collected through mass surveillance using Microsoft Azure, and made the following remarks:

'Azure scans terms within massive piles of text quickly. In this way, even the locations of people giving each other directions can be detected.'

The possibility of artificial intelligence systems making errors was also addressed in the investigation. An Israeli official cited an error in Arabic translations as an example. Microsoft's relationship with the Israeli army has strengthened even further since October 7. According to the AP, Microsoft Azure has responded to 130 technical support requests from the Israeli army in the last 10 months. The company operates a 46,000-square-meter campus in the city of Herzliya, near Tel Aviv.

CIVILIANS IN THE CROSSHAIRS OF ARTIFICIAL INTELLIGENCE

The investigation also revealed that artificial intelligence systems identified approximately 1,000 high school students as 'potential militants.' Officials warned that such errors could lead to civilian casualties.

The official, who spoke on the condition of anonymity, emphasized that the use of artificial intelligence also makes civilians targets, stating the following:

'The Arabic word describing the grip on a rocket-propelled grenade launcher is the same as the expression for "payment." The artificial intelligence translated this incorrectly, and the personnel monitoring it did not initially notice this error.'

Heidy Khlaaf, a former senior official at OpenAI, commented on the research findings, saying, 'This is the first confirmation we have received that commercial artificial intelligence models are being used directly in warfare.'

News Source: AA