Un cap a été franchi : l’IA est au cœur de la guerre au Moyen-Orient

The United States is using AI on an unprecedented scale in Middle East warfare, particularly in Iran and Lebanon bombardments. After 5 days of war, the US military bombed 2000 targets in Iran, including approximately 1000 in the first 24 hours, a level made possible largely through AI use.

• The Pentagon launched Project Maven in 2017 to analyze thousands of hours of satellite, radar, and sensor footage • AI algorithms automatically detect and locate potential targets like individuals and installations • AI assists commanders with weapon supply management and recommends the most effective missiles based on context

AI tools come from tech giants like Amazon and Microsoft, but startups like Palantir and Cloud AI are primarily used by the US to prioritize targets in Iran in real-time.

This massive AI usage is confirmed by statements from high-ranking US officials and multiple journalistic investigations.

Anthropic signed a 200 million dollar contract with the Pentagon after developing Claude AI.

• Dario Amodei, CEO of Anthropic, refused to allow Claude to be used by government for mass surveillance of Americans • Rejected autonomous weapon deployment without human supervision • Insisted on maintaining ethical boundaries in military applications

The Pentagon demanded total access to Claude without exceptions, which intensified disagreements and led to Anthropic being excluded from all sensitive US defense contracts.

Anthropic filed multiple lawsuits arguing the government punished them for defending their AI convictions. The administration gave the military 6 months to stop using Claude, with alternatives like ChatGPT and xAI already positioning themselves as replacements.

The primary criticism from experts and NGOs is that human involvement is decreasing in decisions that can result in death.

• AI targeting systems significantly compress the detection-decision-strike cycle that previously took decades to fully decide and validate by humans • Experts warn that commanders may rely too heavily on AI recommendations without verification • The system can make mistakes that humans might overlook

During Israel's 2023 Gaza offensive, journalistic investigations by The Guardian and other media found that some AI recommendations were validated by humans in mere seconds with no added value beyond rubber-stamping the decision.

A US Army investigation now opens the door to possible American responsibility for bombing an elementary school in Iran that killed 168 people, mostly young girls. The Wall Street Journal reports the US AI program may have relied on outdated intelligence, confusing the school with an old military building that stood there years earlier.

Who is responsible for strikes: the AI system or the person who validated its choice?

The school was located next to a complex composed of multiple buildings linked to the Islamic Revolutionary Guards, the military and ideological branch of the Iranian Islamic regime.

Multiple NGOs are calling for increased regulation and transparency in AI military applications.

The ethical and legal challenges around AI responsibility in warfare are becoming increasingly important, with complex debates about accountability in military strikes likely to continue.

France is tripling humanitarian aid to Lebanon by sending 60 tons of aid to refugees fleeing the south. This includes sanitary and hygiene kits, mattresses, lamps, and a mobile health post. France is also providing armored transport vehicles to Lebanese armed forces.

Karine Buisset, a French humanitarian worker for UNICEF in the Democratic Republic of Congo, was killed in a drone strike on Goma. The exact origin of the attack has not been confirmed and investigations are ongoing.

The Zoro Ranch, a luxury property in New Mexico belonging to sex offender Jeffrey Epstein, was searched on Monday. Justice Department documents indicate bodies may be buried there. A criminal investigation opened on February 19 regarding illegal activities before Epstein's 2019 prison death.

• UN commission accuses Russia of crimes against humanity by forcibly deporting approximately 20,000 Ukrainian children to Russian territory and preventing their return • Bus fire in Kerzers, Switzerland kills six people; one passenger allegedly set the fire deliberately • Deadline for French students to submit university preferences on Parcours Sup is Thursday March 12 at 11:59 PM