General Analysis Raises $10M in Seed Funding to Secure Agentic AI

SAN FRANCISCO--(BUSINESS WIRE)--Apr 29, 2026--

General Analysis, a company building security infrastructure for agentic AI, today announced $10 million in seed funding led by Altos Ventures, with participation from 645 Ventures, Menlo Ventures, Y Combinator, and additional strategic investors and angels. The company was founded by former NVIDIA, Cohere, and DeepMind researchers and is working to close a widening security gap as companies race to put autonomous agents into production. General Analysis is already working with enterprise customers in support and finance whose products and workflows are used by hundreds of millions of users.

This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20260429247972/en/

In March, General Analysis’s adversarial agent convinced 50 live customer service AI agents to give away more than $10 million in fabricated perks — million-dollar gift cards, years of free home security, whatever it could extract — in roughly three minutes per target. Out of 55 bots tested, only five refused. That kind of stress test, along with the defenses it informs, is what General Analysis runs for enterprise customers before their agents reach production.

General Analysis was founded by Rez Havaei, previously an AI researcher at Cohere and NVIDIA, with Maximilian Li, AI safety researcher from Harvard, and Rex Liu, machine learning researcher at Caltech. The trio built the company on the view that securing AI agents is a distinct technical discipline requiring skills and methods different from traditional cybersecurity. Agentic systems behave non-deterministically, and their failures cannot be anticipated by reading code alone.

Across nearly every industry, enterprises are deploying AI agents into increasingly consequential workflows. The upside is large enough that delaying deployment is often not a serious option, and the burden of making these systems safe has largely fallen on security teams. But there is not yet a mature playbook for securing agentic AI.

Last summer, researchers at General Analysis showed that a widely used Supabase integration in Cursor, a code generation AI agent, could allow an internal agent to be hijacked by a single malicious support ticket, tricking it into leaking a complete private database. Simon Willison, the British engineer who coined the term “prompt injection,” cited the finding as a case of the “lethal trifecta” — an AI system that simultaneously holds private data, ingests untrusted content, and can communicate externally.

Security teams are often stuck choosing between agents that are locked down to the point of uselessness and agents whose risks they cannot actually see. “We hear from security teams that they want agents that are secure by design,” said Rez Havaei, CEO of General Analysis. “What that often turns into in practice is a stack of isolation layers and ad hoc context restrictions that makes a system feel more controlled. Those measures either fail to eliminate the underlying vulnerability or constrain the agent enough to limit its usefulness. The problem is that feeling safer and being safer are not the same thing.”

“Our position is that security for AI systems is an empirical problem. It has to be grounded in rigorous measurement of how those systems behave under realistic and adversarial conditions. You cannot prove an agent is safe," said Maximilian Li, co-founder of General Analysis. "You can only measure how often it fails, and how badly, and drive both numbers down."

From that premise, General Analysis aims to help enterprises configure defenses through realistic empirical measurement. Different defensive layers carry different tradeoffs, and there is no universal configuration that makes every agentic system robust. The company combines adversarial evaluations with a broad defensive toolkit to identify the failure modes present in a deployment, measure the effect of different interventions, and help customers determine which configurations materially reduce risk.

“One advantage of agents is that they are much easier to study systematically than the human workflows they are beginning to replace,” said Rex Liu, co-founder of General Analysis. “Many of those workflows were never especially secure to begin with, and their failures are often hard to observe or improve rigorously. But as those workflows become agentic, they also become more measurable and more improvable — which creates a path for businesses to become more secure in practice than they were before.”

“Agentic systems represent a paradigm shift in security. Safety and security in the AI era demand continuous adversarial testing rooted in deep research, not static rule sets,” said Tae Yoon, Partner at Altos Ventures. “Rez, Rex, and Max are exactly the kind of team this moment calls for: technically brilliant, deeply scrappy, and moving incredibly fast. We’re proud to lead this round and partner with them from the earliest days.”

To get in touch with the General Analysis team, reach out at info@generalanalysis.com.

About General Analysis

General Analysis is an AI security company that helps enterprises deploy agentic AI safely through adversarial evaluation and defensive tooling. Founded in 2025 and based in San Francisco, the company is backed by Altos Ventures, 645 Ventures, Menlo Ventures, and Y Combinator. Learn more at generalanalysis.com.

About Altos Ventures

Founded in 1996, Altos Ventures is a technology investment firm based in Silicon Valley. Through patient and pragmatic investing, Altos partners with early to growth stage technology companies operating in consumer and enterprise sectors, with the goal of building durable and compounding businesses over decades. As a registered investment advisor (RIA) with the SEC, Altos has a uniquely flexible, long-term, and concentrated approach to venture capital, supporting the full lifecycle of companies from inception to global growth and profitability. For more information, please visit www.altos.vc.

The SF-based startup, founded by former NVIDIA, Cohere, and DeepMind researchers, is working to close a widening security gap as companies race to put autonomous agents into production

WASHINGTON (AP) — The Supreme Court on Monday rejected President Donald Trump's bid to restrict mail ballots for the midterm elections, capping a flurry of last-minute legal action with voting already underway.

The decision allows states to continue sending out mail ballots under the same processes they’ve used for years and have accounted for nearly a third of votes cast.

It represented a stark defeat for Trump on an issue he consistently has emphasized as vital to ensuring election integrity even though there is virtually no evidence of fraud with mail ballots.

The Supreme Court majority wrote that the administration's push to implement the restrictions this year is likely to lose in court, though the brief emergency order didn't detail its reasoning. Two justices, Samuel Alito and Clarence Thomas, publicly dissented.

The Trump administration had asked the justices, once again at the center of a roiling political controversy, to clear the way for restrictions before the pivotal November contests for control of Congress.

Alito wrote in his dissent that the Postal Service “has broad authority to regulate the mail” and likely does have the power to enforce Trump's restrictions.

Justice Brett Kavanaugh, meanwhile, agreed that the restrictions should not go into effect for the midterms but indicated he might rule in favor of the Trump administration if the issue comes back before the court at a later time.

Election officials have said there was simply no way to carry out a complete overhaul in the weeks before the midterms. Indeed, Alabama, North Carolina and Wisconsin began sending mail ballots to voters over the past week while the new system was still not active.

The Trump administration plan would require states to adopt a uniform envelope style and submit lists of eligible voters to an online portal. The Postal Service could refuse to deliver ballots to states that didn’t comply.

The administration’s efforts had the potential to be especially disruptive in states that send out ballots entirely by mail. Washington's Secretary of State Steve Hobbs, a Democrat, said the decision means work “to carry out a safe, secure, and accurate election" can continue ”without having to upend our election processes to meet unrealistic ballot mail requirements.”

In Arizona, another largely vote-by-mail state, Democratic Secretary of State Adrian Fontes said “it is crucial for us to continue rejecting the notion that access and security are mutually exclusive when it comes to running strong elections.”

Utah Lt. Gov. Deidre Henderson, a Republican serving as the state’s chief election officer, said on social media that the Supreme Court’s decision means “Utahns can have confidence that the 2026 election will proceed as normal."

The White House and Justice Department did not immediately respond to requests for comment, but have in the past framed the restrictions as commonsense changes aimed at keeping elections secure.

Democratic state officials and voting rights groups challenged the restrictions in court, arguing that the president has no constitutional authority to set election rules that would “virtually eliminate mail voting on the eve of a major election."

A whistleblower report, meanwhile, said the postal system’s requirements could lead to millions of mail ballots never being sent, because problems with the not-yet-operational portal meant a single bar code error could result in an entire batch of ballots being tossed out.

Some states don't have the infrastructure to quickly pivot back to in-person voting, according to court documents.

The League of Women Voters, a plaintiff in the case, called the proposed restrictions “an extraordinary attempt to interfere with voting” and said it would have thrown elections into chaos. “Voters should never be caught in the middle of a political power warfare,” said CEO Celina Stewart.

Lower courts agreed to block Trump’s plan, including a preliminary injunction issued by a judge nominated by the president.

But the Trump administration appealed to the Supreme Court, arguing that federal control of the Postal Service allows them to set rules for mail ballot handling and that compliance was possible.

The federal government won an early procedural decision at the Supreme Court, but the justices pointedly did not decide the legality of the plan.

Trump has long opposed mail voting and falsely blamed it for his 2020 election loss to Democrat Joe Biden, even though he often uses that method to cast his own ballot, including as recently as this year.

A report by the Brookings Institution published in 2025 found that mail voting fraud occurred in only about four cases out of every 10 million mail ballots cast.

__

Associated Press writers Nicholas Riccardi in Mason, Michigan; Ali Swenson in New York; Gene Johnson in Seattle; and Jacques Billeaud and Wufei Yu in Phoenix contributed to this report.

FILE - Election worker Wendy Wimer looks through vote-by-mail ballots for the presidential primary election, March 12, 2024, at the Clark County Elections Office in Vancouver, Wash. (AP Photo/Jenny Kane, File)