Cyber criminals increasingly see email as a weak link they can exploit; AI could be the answer
getty
Cybersecurity attacks perpetrated through emails are costing organisations all around the world a fortune. IT giant IBM says phishing emails are now the world’s most common attack vector for data breaches – and that the average breach costs $4.45 million to resolve – but the threat posed by email is much broader. Attackers also use email to launch ransomware attacks, deliver malware and pursue frauds such as spoof executive requests.
Enter Cy Khormaee and Ryan Luo, two former Google executives best known for their work on the reCAPTCHA security service at the internet giant. Their new venture is AegisAI, a start-up focused on using artificial intelligence (AI) to improve email security. The company is today announcing it has raised $13 million of seed funding.
Khormaee argues that the evolution of email security solutions over the past 30 years has been too slow to keep pace with the fast-developing threat – and that with cyber attackers increasingly using AI themselves, the gap is growing dangerously wide.
“Email security started out with a narrow focus on reputational risk, screening out messages from senders deemed potentially untrustworthy, but that was a blunt tool that is easy to circumvent,” says Khormaee. “Since then, we’ve seen the launch of rules-based tools, that look for suspicious emails according to lessons learned from previous attacks, but they’re too slow to keep up with modern email attacks, which are constantly evolving and often bespoke.”
Given these shortcomings, cyber criminals are enjoying increasing success with email attacks, particularly as they make use of new technologies. Research published last year by Crowdstrike found that phishing messages generated by large language models saw a 54% click-through rate, compared to only 12% for human-generated phishing attempts.
AegisAI’s alternative is a solution harnessing AI agents that aim to negate the threat by ensuring suspicious emails never reach the intended recipient. The agents analyse each email received by an organisation and are trained to identify potential red flags – anything from an unusual looking sender address to an urgent request for the recipient to take action.
“Our AI agents are not simple machine learning models, reputation or rule engines; they analyse every email through the lens of intent, context, and user relationships,” Khormaee explains. “Each agent has a specialised function, such as impersonation analysis, financial intent detection or behavioral anomaly detection, and together they simulate how a skilled human analyst would triage a suspicious email – only faster and at scale.”
Based on the development of its platform so far, Khormaee estimates that AegisAI’s tools are 20% more effective at identifying rogue emails than conventional email security platforms. He also claims the platform blocks 90% fewer safe emails – an important advance, given that organisations can’t afford to take a heavy-handed approach to security that potentially stops legitimate and important emails getting through. “We don’t believe in creating more alerts,” adds Luo. “We believe in creating better security outcomes.”
While the company was only founded this year, these successes are beginning to cut through with clients. AegisAI’s early contract wins include Mesh, a San Francisco-based crypto payments specialist, where CEO Bam Azizi says: “As a former security founder, I’ve seen the cat-and-mouse game play out for decades, especially in email security, but Aegis came into Mesh and stopped attackers in their tracks”. More recently, the company has also added data privacy company Lokker, another Californian business, to its client list.
“When will we be able to stop all email attacks?,” asks Khormaee. “The answer, unfortunately, is probably never, but we’re increasingly confident that we can have a huge impact with an agentic solution that makes email safer and more secure than ever before.”
That promise excites investors in the company too. Today’s $13 million seed funding round is led by Accel and Foundation Capital, with AegisAI planning to use the funding to accelerate product development, expand its engineering team, and increase its go-to-market activity.
“The AI era will inevitably drive disruption in email – the easiest attack vector,” says Accel partner Eric Wolford. “Cy and Ryan are both AI-native and have spent an enormous amount of time in email security at Google.”
Still, it’s an increasingly competitive marketplace, with other start-ups also recognising the potential of AI to enhance security. Fundraising rounds in recent months for email-focused security companies have included a $25 million Series A raise for Trustifi, a $20 million Series A round at EasyDMARC, and a $5.2 million seed raise at StrongestLayer, Sublime Security and Abnormal Security have also completed large rounds over the past 12 months.