AI Can Switch DNA Results In Software Used by Many (Most?)Major Forensics Labs

Here is Colorado there has been a lot of attention being paid to a forensics technician who, apparently, took shortcuts with DNA testing. She claims that none of the shortcuts she took changed any results, but she is still awaiting trial on over a hundred charges over it. The big problem is that it may invalidate the evidence of all of the cases that she worked on over a thirty year career working for the state. Details here.

But that is one technician, one state.

What if hackers could have undetectably modified DNA samples or switched them for all DNA tests conducted in all major forensics labs over the last 30 years?

The vendor, of course, says there is no EVIDENCE of that, but if the change is undetectable, there would be no evidence.

They are also WORKING on a fix. Not sure if labs that are using that software have stopped processing samples until that fix is tested and in place or not.

I assume that a reasonably experienced attorney could claim that the DNA used to convict their client is potentially compromised and the defendant is due a new trial. That is what attorneys do – create doubt and exercise the system.

Since the modifications are, apparently, undetectable, I don’t know how the lab could prove beyond a reasonable doubt that the samples were not altered without retesting.

One forensic scientist who was involved in the discovery said “Effectively, what we have are data files that are legitimately referred to as the gold standard of forensic science that lack the same level of tamper-evident markings that we require for a paper bag” (of evidence). That is a bit disappointing.

One more thought. There is no national tracking of how backlogged individual labs are. In Colorado there was a lot attention – very public attention – applied to the backlog of testing just of rape kits. It was 500 days. Colorado has gotten that number down to a bit under 200 days, but consider this:

Lets assume, and it is only a guess, that there were, in the last 30 years, about 1-2 million cases were DNA was a significant factor in the trial and lets assume, again, a guess, that 2 percent of the cases demand a retrial. Further assume that the ORIGINAL DNA sample still exists and is still viable for testing (one guess is that covers about 50 percent of the cases), that would add thousands of DNA tests to the backlog of an already overwhelmed system.

In situations where the original DNA was not preserved, degraded or is not viable for a retest, there is no way to validate that the evidence was not compromised.

Hopefully is doesn’t come to that, but it could be a mess. A big mess. Credit: Tech Radar

by