Helping Solve Crime

Imagine this: A police officer is at a crime scene with no witnesses and no security camera footage. There is no clear motive and no obvious suspects. The only evidence they can find is a bit of DNA in the form of blood or hair. That DNA is compared against a database but there are no matches. No other leads can be found, and eventually the case goes cold. 

Enter Corsight AI, a subsidiary of leading Israeli AI developer Cortica. Corsight AI is developing an AI model that aims to help solve these kinds of situations. Despite a healthy amount of skepticism from scientists, Corsight AI claims this “DNA to Face” technology will be able to analyze a DNA sample and create a model of a face that can then be run through facial recognition systems. Revealed during the Imperial Capital Investors Conference in New York last December, the facial recognition tool “constructs a physical profile by analyzing genetic material collected in a DNA sample.” During the conference, Corsight AI shared a roadmap that highlighted their intentions to expand the recognition capabilities of the AI to include “voice to face” and “movement.” MIT Technology Review pressed them further for details, but were told “we are not engaging with the press at the moment as the details of what we are doing are company confidential.”

So far, the company’s promotional and marketing materials are showing a hard lean towards law enforcement applications. Further reinforcing this is the fact that Corsight AI’s board of advisors includes James Woolsey, former director of the CIA, and Oliver Revell, former assistant director of the FBI. 

According to experts, the science and technology to fully support this system is still quite a ways off, so unless Corsight AI has privately made some major advancements it’s likely we won’t see this kind of technology implemented for years. Some experts have warned of the moral implications of actually using this kind of facial recognition. There are numerous ethical and privacy concerns, particularly surrounding the biases that facial recognition technology already suffers from. It is also a Big Brotherish reminder of the industry’s ambitions and intentions for the future. 

Corsight AI is not the first company to attempt this kind of facial recognition technology. Human Longevity and Parabon NanoLabs have separately developed their own recognition systems with varying degrees of success. According to Parabon’s website, their software helps solve an average of one case per week and has helped close over 600 cases in the last seven years alone. Their system still has some major flaws, such as not being able to guess the shape of or distance between certain facial features, but it is still accurate enough to be useful for law enforcement. 

This is not the first time that incomplete data or hand-drawn sketches have been used with facial recognition systems. In fact, a 2019 study by Georgetown Law’s Center found that at least six different law enforcement agencies in the United States “permit, if not encourage” using either hand-drawn or computer generated forensic sketches to discover potential suspects. Some AI experts have pushed back, warning that this will ultimately lead to overall lower levels of accuracy. 

Corsight has lauded their technology as the “most ethical facial recognition system for highly challenging conditions,” according to a presentation made available online. CEO Robert Watts claims their facial recognition software can “identify someone with a face mask—not just with a face mask, but with a ski mask.” How effective this technology will prove to be in real-life applications remains to be seen, but Corsight is optimistic enough that it began courting investors during December’s conference. This sends clear signals that they are confident in their technology and have made enough progress to move towards the development’s endgame. 

Problems with facial recognition and tracking technology have been well-documented in the past. Those problems are exacerbated when analyzing photographs taken in poor lighting conditions or at extreme or awkward angles. There are also issues when analyzing photos of subjects that are either very young or very old. From a privacy standpoint, there has been criticism leveled at certain facial recognition technology, particularly systems such as Clearview AI that scrape social media pages to improve its processes and make potential matches. 

For some, the idea that this kind of facial recognition technology is anywhere near ready for deployment is utterly out of the question. “The idea that you’re going to be able to create something with the level of granularity and fidelity that’s necessary to run a face match search—to me, that’s preposterous,” claims Albert Fox Cahn, civil rights lawyer and executive director of Surveillance Technology Oversight Project (STOP). “That is pseudoscience.”

Other experts have weighed in, casting doubts on Corsight AI’s claims. Dzemila Sero, a top researcher at Dutch research center Computational Imaging Group, claims the science and technology to support these kinds of systems simply doesn’t exist yet; at least not in the hands of any public company. Citing studies from Human Longevity, she says that the gene catalogs and databases are simply incomplete and therefore incapable of rendering any kind of reliable facial recognition. Additionally, there are factors such as the environment and the aging process that currently cannot be accounted for. In an email to MIT Technology Review, she states “Premature attempts to implement this technique would likely undermine trust and support for genomic research and garner no societal benefit.”

The Future of Facial Recognition Technology

Despite any practicality or ethical concerns, the market for facial recognition technology continues to grow as companies scramble for customers. Corsight AI is but the latest in a long line of businesses offering this kind of technology with flashy promotional material and promises of cutting-edge features. Russian-based NtechLab provides a recognition service that expands upon facial matching to also include body-shape matching as well as license plate reading. This is a far more comprehensive approach in which facial recognition becomes only one facet of the identification process. While Corsight AI’s DNA to Face software clearly indicates their intentions of delving into the world of facial recognition technology, there needs to be some kind of oversight to help mitigate instances of bias or misidentification. Because they have held their cards so close to their chest, it remains unclear whether or not they even have a functioning prototype that exists using today’s technology. If they do have a working model, we would be entering uncharted territory in terms of personal privacy and security.