Deepfake threats spawn new business for entrepreneurs, investors

• Major companies — and even the U.S. presidential election — are prime targets for attackers using these tools.

Driving the news: Norwest Venture Partners published a market map Tuesday of deepfake and brand security vendors that attempts to organize the growing number of companies operating in this space.

• The market map, first shared with Axios, divides the ecosystem into six buckets: image and video detection, audio detection, content watermarking, identity verification, brand protection and content licensing, and narrative attack protection.

The intrigue: Investors see deepfake detection becoming just as necessary as email and application security tools, and startups are primed to lead in this space.

• Model makers like OpenAI and Anthropic are mostly focused on building new consumer and enterprise products for their large language models.
• And traditional cybersecurity vendors would have to build an entirely new tool with biometric data to detect deepfakes, Dave Zilberman, a general partner at Norwest, told Axios.

Catch up quick: Deepfakes used to be obvious and clunky, with audio glitches and poorly edited video.

• But new generative AI tools have made it much easier and faster for malicious actors to refine these fakes — and these tools are improving at a rapid clip.
• In a study published Monday, researchers at Utah Valley University found that 56% of their 244 survey participants believed a deepfake video was real. Half of them also believed a deepfake audio clip was legitimate.

The big picture: Norwest identified a variety of earlier-stage companies that are leading in this space.

• Reality Defender, GetReal Labs and Deed Media are some of the top players in image and video detection, while Pindrop and Resemble AI are among those leading in audio detection.
• Polyguard, Deduce, Nametag and Kibu are new players to watch in the identity verification sector.
• DeepTrust has become a leader in watermarking audio content. EchoMark, Truepic and Steg.AI are leaders in video or document watermarking.

Between the lines: Investors are starting to pour money into this space.

• Reality Defender raised a $33 million Series A round last week. Clarity, a video deepfake detector, raised a $16 million seed round in February.
• Meanwhile, Pindrop secured a $100 million loan from Hercules Capital in July.

Zoom in: Vijay Balasubramaniyan, co-founder and CEO of Pindrop, told Axios that his company has signed 16 customers since releasing its deepfake audio detection tool in February.

• "In six months, we've sold more deepfake detection than our first product, which put us on the map, in [its] first five years," Balasubramaniyan said.
• Pindrop's customers include financial institutions, media companies and communications firms.
• Pindrop went from seeing one deepfake per month across its entire customer base last year to seeing one deepfake per day per customer this year, Balasubramaniyan said.

Threat level: For years, customers bought deepfake detection technology as a kind of insurance policy in case, one day, they saw such a threat.

• Buying security tools with an insurance mindset doesn't last in a tight economy, Zilberman said.
• However, the threat is now real and visceral, Balasubramaniyan noted.
• "We are really seeing a massive explosion of attackers using generative AI and deepfakes to complete their attacks," Balasubramaniyan said. "It is actually a real threat."