Outtake launches Recon Agent to trace AI attacks early

Outtake has launched Recon Agent, a threat investigation tool designed to trace digital attacks before contact with targets begins. It also introduced the Digital Trust Kill Chain, a framework that sets out eight stages of AI-driven attacks.

The New York-based cyber security company said both products are aimed at identity-based threats that use automated systems across email, social media, messaging apps, and spoofed websites. Recon Agent operates at the reconnaissance and infrastructure setup stages, earlier than many existing security tools that focus on engagement or credential theft.

Founded in 2023, Outtake says it is valued at more than USD $100 million. It also announced a USD $40 million Series B funding round led by Iconiq, with backing from Microsoft Chief Executive Officer Satya Nadella and Palo Alto Networks Chief Executive Officer Nikesh Arora.

Threat framework

Modelled on the MITRE ATT&CK framework, the Digital Trust Kill Chain outlines what Outtake describes as eight repeatable stages in modern AI-led attacks: reconnaissance, infrastructure setup, trust exploitation, target engagement, credential capture, account takeover, impact and fraud, and monetisation.

According to Outtake, many defensive tools do not become useful until the fourth stage, when attackers begin direct contact through phishing emails, social media messages, WhatsApp outreach, or fake online events. Recon Agent is intended to intervene earlier by tracing lookalike domains, fake accounts, bot networks, and other assets prepared before an attack is launched.

Outtake said the product is a long-running autonomous agent built with Anthropic. Starting from a single threat signal, the system maps related domains, accounts, pre-staged assets, and campaigns linked to an operator.

The company said this process can reduce investigations that once took weeks to hours. Each investigation also improves the speed of later work by adding intelligence that can be reused.

Early focus

The launch comes as security groups assess how generative AI is changing online fraud and impersonation. Outtake cited findings from its Outtake Labs 2026 Digital Trust Industry Pain Report, which found that 67% of security leaders believe generative AI has materially expanded their attack surface.

Outtake argues that the main weakness for defenders is not a lack of alerts, but the difficulty of acting before attackers adapt and relaunch campaigns. That view reflects a wider cyber security debate over whether automated threats are moving faster than teams that still rely on manual investigations.

Recon Agent joins Outtake's existing products, Search Agent, Triage Agent, and Remediation Agent. Together, they form what the company describes as an agent-led approach to detecting, investigating, and disrupting identity-based threats.

Customers and partners named by Outtake include Anthropic, Square, and AppLovin. Recon Agent is now generally available to all customers.

Alex Arjun Dhillon outlined the company's view of the shift in online threats.

"The agentic internet is not coming. It is here," said Alex Arjun Dhillon, Co-Founder and Chief Executive Officer, Outtake. "As AI agents become the dominant force on the internet, adversaries are already weaponizing them. Recon Agent was built for this moment. It does not remove the fake profile. It removes the operator running hundreds of them."