- The power of combining large databases and biometric data makes the field ‘ripe for AI,’ says University of Kent’s Robert Green
- AI-generated audio and video threaten the reliability of evidence, warns Russian forensic audio expert Ivan Siparov
- - ‘AI should always serve as an aid to human expertise, not as a substitute for it,’ says Sebastiano Battiato of Italy’s Catania University
ANTALYA
Artificial intelligence is reshaping industries across the globe, and forensic science is no exception. At once groundbreaking and controversial, AI is now entering the world of crime-solving, where its potential is as vast as the challenges it raises.
“There are only three ways in which you can ever solve a murder,” said Robert Green, a professor and forensic scientist at the University of Kent in the UK. “One is by eyewitness testimony, one is by confession evidence and the rest is by science and technology. And if you look at the former … it tends not to be believed anymore.”
With the increasing importance of technology, AI is stepping up as a “real game changer,” particularly in fingerprint and DNA analysis, he told Anadolu at the 20th International Forensic Medicine Days in the Turkish resort city of Antalya, which gathered nearly 800 experts from 27 countries.
Advances in forensic technology already allow scientists to obtain more information from increasingly tiny samples, Green explained.
For example, modern DNA analysis can start with just 500 picograms (around 18 trillionths of an ounce), an amount so small it could be found on a watch strap or glasses and is invisible to the naked eye.
But interpreting such data is a different story.
Green likened the challenge to “throwing two jigsaw puzzles into the air and trying to sort out the pieces.”
Analyzing mixed DNA samples requires vast computational power and precision. The power of combining large databases and biometric data makes the field “ripe for AI,” he said.
Especially in traumatic cases, analyzing and recovering mixed DNA is difficult, and any help is valuable, he said.
He pointed to major improvements in fingerprint matching as proof that AI is already delivering results.
Still, Green acknowledged the ethical and legal challenges with AI, saying skepticism persists because few understand how the technology works, and courts will likely remain cautious until AI processes can be fully explained.
The AI arms race
If AI promises breakthroughs, it also poses new dangers. Russian forensic audio expert Ivan Siparov warned that AI-generated audio could undermine the very foundation of evidence.
“This is a threat to society today, and everyone understands that,” he said.
He pushed back against the idea of AI as a silver bullet for forensic science. “Miracles are not possible,” he said.
For him, the integrity of original material is paramount. Now, once a video or audio file is broadcast, re-recorded, or pulled from social media, it loses its evidentiary value. “We need the device itself … We need the trace from the basics,” he said.
He described the escalating contest between technology developers and investigators as nothing short of an “arms race.”
“You have an AI generator and you have an AI discriminator. To catch AI, you need your own AI. AI against AI,” he said.
Another tool, not a replacement
Even Siparov admits AI can help, particularly in the bulk processing of forensic material. Automated systems could flag relevant data and lighten workloads.
Yet, he warned, these tools remain far from infallible, operating as “black boxes” that require carefully designed human countermeasures.
Sebastiano Battiato, a professor of computer science at Italy’s Catania University, echoed both the hope and the hesitation.
AI “has the potential to transform forensic science” in tasks like deepfake detection and digital trace reconstruction, he told Anadolu.
But he cautioned against over-reliance.
“AI should always serve as an aid to human expertise, not as a substitute for it,” he said, urging justice systems to find balance in terms of innovation and regulation.
