75 million deepfakes blocked: Persona leads the corporate fight against hiring fraud

While remote work has become the standard, a dark threat appeared in corporate job services: false candidates sophisticated to AI which can pass video interviews, submit convincing curriculum vitae and even deceive human resources professionals to offer them jobs.

Now, companies rush to deploy advanced identity verification technologies to combat what security experts describe as a growing crisis in candidates, largely motivated by generative AI tools and coordinated efforts of foreign actors, including groups sponsored by the North Korean State seeking to infiltrate American companies.

Based in San Francisco CharacterA leading identity verification platform, a major expansion of its workforce about the workforce, introduced new tools specially designed to detect the characters generated by the AI ​​and deep attacks during the job process on Tuesday. The improved solution fits directly into the main corporate platforms, in particular Okta identity identity cloud And Cisco duoallowing organizations to verify the identities of candidates in real time.

“In today’s environment, ensuring that the person behind the screen is what they claim to be more important than ever,” said Rick Song, CEO and co-founder of CharacterIn an exclusive interview with Venturebeat. “With the actors sponsored by the State infiltrating companies and the identity generation of AI easier than ever, our improved solution for the IDV of the workforce gives organizations the confidence that each attempt at access is linked to a real and verified person.”

The moment of Persona’s announcement reflects an increasing emergency around what cybersecurity professionals call a “identity crisis” in distance hiring. According to a Gartner’s report April 2025By 2028, one in four candidates in the world will be false – an amazing prediction that emphasizes how AI tools have lowered the obstacles to the creation of convincing false identities.

75 million blocked depth attempts reveal a massive range of fraud at hiring fueled by AI

The threat extends far beyond the bad individual actors. In 2024 only, Persona blocked more than 75 million OPENAI,, Course,, InstacartAnd Twilio. The company has observed a 50 -time increase in deep activity in recent years, attackers deploying increasingly sophisticated techniques.

“The threat of North Korean IT workers is real,” said Song. “But it is not only North Korea. Many foreign actors all do things like this with regard to the means of infiltrating organizations. The threat of initiate for companies is higher than ever.”

High -level recent cases have highlighted the severity of the problem. In 2024, cybersecurity company Knowbe4 inadvertently hired a North Korean IT worker Who tried to load malware on business systems. Other fortune companies have been victims of similar regimes, where foreign players use false identities to access sensitive business systems and intellectual property.

THE Department of Internal Security warned that these “deep identities” represent a growing threat to national security, with malicious actors using characters generated by AI to “create videos, images, audio and credible and credible text of events that have never occurred”.

How the three -layer detection technology is fighting against the false plans of sophisticated candidates

Song’s approach to fight against fraud generated by AI is based on what he calls a “multimodal” strategy that examines identity verification on three distinct layers: entry itself (photos, videos, documents), environmental context (device characteristics, network signals, capture methods) and population models that could indicate coordinated attacks.

“There is no miracle solution to really resolve the identity,” said Song. “You cannot look at it from a single methodology. The AI ​​can generate very convincing content if you only look at the level of submission, but all the other parts of the creation of a convincing false identity are always difficult.”

For example, although an AI system can create a false photorealistic head, it becomes much more difficult to simultaneously browse the fingerprints of the device, the characteristics of the network and the behavioral models that personality systems monitor. “If your geolocation is deactivated, then time zones are disabled, time zones are disabled, then your environmental signals are deactivated,” said Song. “All of these things have to go into a single framework.”

The company detection algorithms are currently outcoming humans to identify Deepfakes, although Song recognizes that it is a arms race. “The AI ​​improves better and better, improving more quickly than our ability to detect only the level of entry,” he said. “But we look at the progression and adaptation of our models accordingly.”

Business customers deploy the verification of the identity of the workforce in less than an hour

The improved labor check solution can be remarkably deployed quickly, according to Song. Organizations are already using Okta Or Cisco Identity management platforms can integrate Persona’s screening tools as little as 30 minutes at an hour. “The integration is incredibly fast,” said Song, crediting the OKTA team for having created transparent connectivity.

For companies concerned about user experience, Song stressed that legitimate candidates generally complete the verification in a few seconds. The system is designed to create “friction for bad users to prevent them from passing” while maintaining a fluid experience for authentic candidates.

Large technological companies already see results. OPENAIwhich deals with millions of user checks each month via Persona, reaches automated screening of 99% with only 18 milliseconds of latency. The IA company uses the screening capacities of persona sanctions to prevent bad actors from accessing its powerful language models while maintaining an experience of registration without friction for legitimate users.

The market for verification of the identity pivots from the checking of the history of proof candidates exist

The rapid adoption of hiring fraud fueled by AI has created a new market category for identity verification specifically adapted to labor management. Companies for verifying traditional history, which check the information on candidates, after having supposed that their identity is authentic, is not equipped to deal with the fundamental question of knowing if a candidate is what they claim to be.

“The history checks assume that you are what you say, but then check the information you provide,” said Song. “The new problem is: are you who tell you that you are?” And it is very different from what the lapart checking companies traditionally resolve. ”

Change to remote work has eliminated numerous traditional identity verification mechanisms. “You have never had a problem knowing that if someone presents himself in person, you know with a relatively high certainty, you are who you say that you are,” noted Song. “But if you interview by zoom, all this could be a Farfake Deep.”

Industry analysts expect the market to verify the identity of the workforce is developing rapidly because more and more organizations recognize the scope of the threat. According to Markets and marketsThe global identity verification market is expected to reach $ 21.8 billion by 2028, compared to $ 10.9 billion in 2023, representing an annual growth rate made up of 14.9%, with labor applications representing one of the fastest growth segments.

Beyond Deepfakes detection: the future of digital identity lies in the history of behavior

While the technological arms race between the fraud and detection systems generated by AI is intensifying, Song thinks that the ultimate solution may require a fundamental change in the way we think of the verification of identity. Rather than focusing solely on detecting if the content is artificially generated, it envisages a future where digital identity is established thanks to an accumulated behavioral history.

“Perhaps the long-term question is really not whether it is AI or not, but really that is responsible for this interaction,” said Song. The company explores the systems where identity would be proven thanks to the digital imprint of a person – its history of legitimate transactions, complements of courses, purchases and verified interactions on several platforms over time.

“All the previous actions that I have accomplished – Commander in Doordash, ending a course on Racera, buying shoes in Stockx – these long -term interactions are probably those that will really define who I am,” said Song. This approach would make exponentially more difficult for bad players to create convincing false identities, because they would need to make years of authentic digital history rather than a simple convincing video or document.

Strengthening Person Idv of the workforce The solution is available immediately, with support for the verification of the identification government in more than 200 countries and territories, and integration capacities with the main identity and access management platforms. While the Revolution of Distance Works continues to reshape the operation of companies, companies are found in an unexpected position: having to prove that their candidates are real people even before they can start to check their qualifications.

In the digital age, it seems, the first qualification for any job can simply exist.