A look at efforts to scrutinize TrueAllele's secret algorithm, which US prosecutors are using to assess the likelihood of a suspect's DNA in a mixed sample
Justin Jouvenal / Washington Post : Tweets: @docedge85 , @maxcembalest , and @brookingsgov Tweets: Doc Edge / @docedge85 : As for TrueAllele, 3 judges have granted review permission recently under heavy protective orders, but the reviews haven't happened yet. In one case the prosecutors withdrew evidence rather than going through the (no doubt onerous) process (14/) https://www.washingtonpost.com/ ... Max Cembalest / @maxcembalest : “We shouldn't be using the criminal justice system as a proving ground for new technologies, especially when the makers of these technologies are keeping how they work secret.” https://www.washingtonpost.com/ ... Brookings Governance / @brookingsgov : “The whole justice system is becoming automated,” @RebeccaWexler told @WashPost. https://www.washingtonpost.com/ ...
Context & Ripple Effects
The Washington Post's look at courts relying on proprietary third-party algorithms lands at a hinge moment for TrueAllele: three judges have granted permission to review the DNA-analysis code, but only under heavy protective orders, and no completed review has yet emerged from those grants.
The pattern is familiar from adjacent coverage — defense lawyers have mounted similar accuracy and reliability challenges against Cybercheck, another black-box tool used across thousands of US cases — making TrueAllele a test of whether trade-secret forensics can survive adversarial scrutiny at all.
First-order effects
- Defendants in mixed-sample DNA cases now face an evidentiary fork shaped by secrecy: judges can authorize protected source-code reviews of TrueAllele, but the process is onerous enough that in one case prosecutors withdrew the evidence rather than undergo it.
Second-order effects
- Every withdrawal or stalled protective-order review raises the litigation cost of using TrueAllele, pressuring prosecutors to weigh secret-algorithm evidence against cases they might lose on disclosure grounds — while vendors face pressure to open their code to keep court admissibility.
- Critics like Max Cembalest are framing the courtroom itself as an improper proving ground, which pushes the debate toward pre-adoption validation standards rather than case-by-case defense challenges.
Third-order effects
- If protective-order review becomes the standard mechanism, forensic software vendors will operate under a de facto audit regime where admissibility depends on granting limited code access — a structural shift from 'trust the vendor' toward verifiable-but-confidential evidence tools, echoing Rebecca Wexler's warning that the justice system as a whole is becoming automated with little visibility into how.
The trend: US courts are converging on judge-mediated, protective-order audits as the only workable compromise between trade-secret forensic AI and defendants' ability to challenge it.