The CIA Let AI Write Its First Intelligence Report—And AI 'Coworkers' Are Up Subsequent – Decrypt

The CIA just lately used AI to generate an intelligence report with no human analyst driving it. Deputy Director Michael Ellis confirmed the milestone Thursday at a Particular Aggressive Research Mission occasion, marking a shift from quiet experimentation to a public declaration of ambition.

Ellis stated the company ran greater than 300 AI tasks final yr, Politico reviews. Someplace in that stack, a machine produced an intelligence product completely by itself—a primary within the company’s historical past.

The near-term roadmap is extra incremental. Analysts would get AI “coworkers” embedded in company analytics platforms to deal with drafting, enhancing for readability, and benchmarking outputs in opposition to tradecraft requirements. People would nonetheless finally sign-off on the outcomes. However the aim is pace—getting intelligence merchandise out sooner than a human-only pipeline permits.

Inside a decade, Ellis stated, CIA officers will handle groups of AI brokers working as “autonomous mission companions,” a hybrid mannequin that scales intelligence gathering in methods no human workforce can match alone.

The CIA has been constructing towards this for years. In 2023, the intelligence company introduced its personal AI chatbot to assist staffers parse surveillance information. By 2024, CIA Director Invoice Burns and MI6 Chief Richard Moore collectively disclosed they had been actively utilizing generative AI for content material triage, analyst help, and monitoring how overseas adversaries deploy the know-how. Ellis’ remarks push that public timeline ahead significantly.

Earlier this yr, Anthropic declined to loosen up restrictions barring its instruments from home surveillance or absolutely autonomous weapons purposes. Protection Secretary Pete Hegseth responded by designating Anthropic’s merchandise a “provide chain danger.” President Trump then ordered each federal company to part out Anthropic instruments. The corporate has legally challenged the transfer.

Ellis did not identify Anthropic, however the message landed clearly. The CIA “can not permit the whims of a single firm” to constrain its use of AI, he stated, and the company is actively diversifying throughout distributors to remain operationally versatile.

Ellis additionally flagged that the CIA doubled its technology-focused overseas intelligence reporting, monitoring how adversaries like China are deploying AI throughout semiconductors, cloud computing, and R&D. The company’s Heart for Cyber Intelligence was elevated to a full mission heart—a transfer Ellis described as important, on condition that “the battle of cybersecurity will likely be a battle of synthetic intelligence.”