The Deepfake Detection Arms Race: Startups Battle to Authenticate Content in Israel’s War

1. Advanced deepfake technology is enabling the spread of misinformation during wars and conflicts, sowing confusion and potentially impacting real-world violence.
2. Startups and tech companies are racing to build deepfake detection tools to authenticate content, but generative AI’s rapid pace may allow malicious actors to outmaneuver them.
3. Analyzing the flood of graphic and traumatic images circulating online takes an immense emotional toll on the teams working to separate real footage from fakes.

As dawn broke on Oct. 7, air-raid sirens blasted out across Tel Aviv, alerting residents to take shelter from an apparent attack. Messages on cell phones soon revealed that Hamas gunmen were allegedly waging mass slaughter on Israelis and seizing hundreds of hostages less than an hour away. Over 1,200 Israelis were reported dead, and the country was seemingly at war.

Michael Matias, CEO of the AI startup Clarity which detects deepfakes, immediately assembled an emergency meeting with his team that morning. They determined that their technology would be valuable in authenticating the flood of violent images spreading online about the attacks.

Deepfake Videos Flood Social Media in Israel-Gaza War: AI Startups Rush to Detect Manipulated Content

In the aftermath, Israel blocked supplies to Gaza and bombed what it claimed to be Hamas targets in the densely populated Palestinian coastal enclave. As the crisis worsened with over 10,000 Palestinians killed, graphic images circulated globally, igniting protests and rage. Along with the outrage, questions emerged over whether some gruesome scenes were even real or AI-generated deepfakes.

Advanced deepfakes have become increasingly realistic and difficult to detect. Startups like Clarity are engaged in a “cat and mouse game” to identify manipulated videos before they cause further violence and confusion. The stakes are high as generative AI tools become more accessible to malicious entities. Governments and companies are struggling to keep pace with the technology.

Since Russia’s invasion of Ukraine in February 2022, fake videos have emerged of both President Zelensky and President Putin calling for their troops to surrender, exposing the potential impact of improved deepfakes. Now widely available AI chatbots like ChatGPT have enabled average users to easily create manipulated content. The consequences in war can be severe.

Both sides in the Israel-Gaza war have spread deepfakes, often using fake images and videos found online rather than AI-generated ones. The flood of questionable information online has made finding the truth difficult.

Matias’ experience in Israeli intelligence units informed his founding of Clarity to solve this problem. As media outlets emailed Clarity for assistance in evaluating suspect videos and images, the team began meticulously analyzing hundreds of gruesome war images. Rather than outright labeling content as real or fake, Clarity uses AI to assess multiple data points and assign a certainty rating. Some videos are clear manipulations but others require more nuanced human judgement.

Clarity is not the only company working to detect deepfakes. Several startups and tech firms have recently launched similar initiatives. Governments are also responding but critics argue their efforts lack teeth. Regardless, there is growing consensus that the accelerating progress of generative AI poses catastrophic societal risks if left unchecked.

The emotional toll on Clarity’s team of reviewing traumatic images is palpable despite the sense of purpose in their work. The stakes of deepfakes continue to rise as U.S. warships patrol nearby and other regional powers remain on alert. For Matias’ team and startups like Clarity, the deepfake detection arms race is just beginning.

The proliferation of deepfakes has opened a Pandora’s box of misinformation with sobering real-world consequences in conflict zones. The fog of war now includes the viral spread of manipulated media alongside actual battlefield footage. This development challenges the very notion of an “objective truth” in wartime.

In Israel’s war with Gaza, the stakes are especially high as U.S. warships stand ready offshore with other regional powers on alert. The accused perpetrators of violence vociferously deny such allegations, instead pointing to deepfakes as the true culprit. Startups race to separate falsified evidence from the legitimate, but their solutions rely on AI technology that pales in sophistication compared to what they aim to defeat.

Those on the front lines endure immense psychological duress, immersed daily in streams of graphic content as they labor to protect the integrity of public discourse. However, the sheer volume of media generated by modern connectivity far outpaces their capacity. Though some fakes may be swiftly debunked, a residue of doubt remains regarding all evidence.

This corrosive effect serves to advantage the cunning who deal in deception and propaganda. When the authenticity of any image or sound can be called into question, achieving public consensus on basic facts becomes improbable. Without an established baseline reality, fruitful debate shuts down and passion-driven rhetoric fills the void.

The emergence of deepfake technology has thus stripped away the guardrails in public forums. Uncertainty breeds suspicion, division, and hostility. Until generative AI can be safely caged, its disruptive potential appears likely to undermine social cohesion and empower bad actors during times of conflict. The turbulent years ahead will test whether truth and trust can endure in the digital wilderness.