Skip to Content
logologo
AI Incident Database
Open TwitterOpen RSS FeedOpen FacebookOpen LinkedInOpen GitHub
Open Menu
Discover
Submit
  • Welcome to the AIID
  • Discover Incidents
  • Spatial View
  • Table View
  • List view
  • Entities
  • Taxonomies
  • Submit Incident Reports
  • Submission Leaderboard
  • Blog
  • AI News Digest
  • Risk Checklists
  • Random Incident
  • Sign Up
Collapse
Discover
Submit
  • Welcome to the AIID
  • Discover Incidents
  • Spatial View
  • Table View
  • List view
  • Entities
  • Taxonomies
  • Submit Incident Reports
  • Submission Leaderboard
  • Blog
  • AI News Digest
  • Risk Checklists
  • Random Incident
  • Sign Up
Collapse

Incident 823: Cybercheck Tool Allegedly Provides Questionable Evidence in Murder Trials

Description: Global Intelligence's Cybercheck AI tool, used by law enforcement to track suspects based on open source data, has allegedly been providing inaccurate or unverifiable evidence in several murder trials. Reportedly the tool lacks transparency and often produces unreliable reports, which has prompted prosecutors to withdraw Cybercheck evidence from multiple cases after its findings were challenged, reportedly wasting law enforcement time and resources while undermining prosecutors' cases.
Editor Notes: This incident ID is a collective incident representing several discrete harm events centering on investigations into Global Intelligence's Cybercheck AI tool, an incident ID that may form the basis for future individual incident IDs themselves. Quick notes on the individual incidents and their general timelines for later consideration: (1) In Akron, Ohio, on August 2, 2020, Kimberly Thompson, Brian James, and Thompson’s grandson, Tyree Halsell, were shot. Cybercheck evidence mistakenly placed suspect Phillip Mendoza at the scene on August 20, claiming 93.13% accuracy. A corrected report later appeared with the correct date, but questions arose over its accuracy. Mendoza’s defense challenged the evidence, and the Cybercheck report was ultimately withdrawn. Mendoza pleaded guilty in August 2023, receiving 15 to 20.5 years, though Cybercheck’s errors may have influenced the outcome. (2) From 2017-2021, Cybercheck expanded rapidly across U.S. law enforcement agencies, conducting thousands of searches, often without methodological validation. (3) In Midland County, Texas, 2021, Cybercheck linked suspect Sergio Cerna to a murder scene by claiming a 97.25% accuracy rate in his device's interaction with a wireless printer. Mid-hearing, the prosecution withdrew Cybercheck’s evidence due to a lack of corroboration, yet Cerna was later convicted on other grounds. (4) In July 2021, in Akron, Ohio, Cybercheck implicated Deshawn Coleman and Eric Farrey in a robbery. However, experts could not verify social media accounts or email addresses tied to the suspects. (5) In 2022, in Aurora, Colorado, Cybercheck identified a homicide suspect via a router ping, but detectives found the resident had no such router, nullifying the evidence. (6) In a separate Akron 2021 homicide, Cybercheck data against Javion Rankin was excluded after Adam Mosher withheld the source code, leading to Rankin’s release. (7) In another Aurora, Colorado case involving the 2022 shooting of a 13-year-old, Cybercheck posited a gang link between rival suspects, conflicting with other evidence and raising doubts. (8) Mississippi’s Bureau of Investigation, in 2022, used Cybercheck in a missing juvenile case but found the information false; the juvenile was found through traditional methods, reportedly wasting investigative resources. (9) In late 2022/early 2023, Summit County prosecutors withdrew Cybercheck evidence from four murder cases following defense challenges on reliability. (10) In a New York cold case, Cybercheck claimed to track a suspect’s phone to multiple crime locations from 20 years ago, with trial set for 2025. (11) In late 2022, Summit County murder cases for Salah Mahdi and Adarus Black involved Cybercheck, but no defense challenge was raised. Later, validation issues in other cases prompted post-conviction concerns. (12) In a 2023 Boulder, Colorado child sexual abuse case, Mosher’s reportedly misleading testimony about prior Canadian cases led to dropped charges and raised credibility questions. (13) In April 2023, Coleman and Farrey’s defense revealed inconsistencies in Mosher’s expert witness records, undermining the Cybercheck report. (14) Ohio’s Bureau of Criminal Investigation ended its $30,000 contract after Cybercheck provided unverifiable email leads in multiple cases. (15) Yakima County, Washington ceased using Cybercheck following numerous inaccuracies in the tool’s output. (16) Cybercheck’s reports included persistent issues with unverifiable device-network pings, casting doubts on the legitimacy of its claimed open-source accuracy.

Tools

New ReportNew ReportNew ResponseNew ResponseDiscoverDiscoverView HistoryView History

Entities

View all entities
Alleged: Global Intelligence and Cybercheck developed and deployed an AI system, which harmed Phillip Mendoza , Sergio Cerna , Unnamed Aurora Colorado residents , Mississippi Bureau of Investigation , Four unnamed Summit County Ohio men , Unnamed Boulder County Colorado resident , Ohio Bureau of Criminal Investigation and Yakima County Sheriff’s Office.

Incident Stats

Incident ID
823
Report Count
3
Incident Date
2024-05-03
Editors
Applied Taxonomies
MIT

MIT Taxonomy Classifications

Machine-Classified
Taxonomy Details

Risk Subdomain

A further 23 subdomains create an accessible and understandable classification of hazards and harms associated with AI
 

7.4. Lack of transparency or interpretability

Risk Domain

The Domain Taxonomy of AI Risks classifies risks into seven AI risk domains: (1) Discrimination & toxicity, (2) Privacy & security, (3) Misinformation, (4) Malicious actors & misuse, (5) Human-computer interaction, (6) Socioeconomic & environmental harms, and (7) AI system safety, failures & limitations.
 
  1. AI system safety, failures, and limitations

Entity

Which, if any, entity is presented as the main cause of the risk
 

Human

Timing

The stage in the AI lifecycle at which the risk is presented as occurring
 

Post-deployment

Intent

Whether the risk is presented as occurring as an expected or unexpected outcome from pursuing a goal
 

Intentional

Incident Reports

Reports Timeline

+1
An AI tool used in thousands of criminal cases is facing legal challenges
Questions arise over the use of an AI crime-fighting toolThis AI Tool Helped Convict People of Murder. Then Someone Took a Closer Look
An AI tool used in thousands of criminal cases is facing legal challenges

An AI tool used in thousands of criminal cases is facing legal challenges

nbcnews.com

Questions arise over the use of an AI crime-fighting tool

Questions arise over the use of an AI crime-fighting tool

theweek.com

This AI Tool Helped Convict People of Murder. Then Someone Took a Closer Look

This AI Tool Helped Convict People of Murder. Then Someone Took a Closer Look

wired.com

An AI tool used in thousands of criminal cases is facing legal challenges
nbcnews.com · 2024

Law enforcement agencies and prosecutors from Colorado to New York have turned to a little-known artificial intelligence tool in recent years to help investigate, charge and convict suspects accused of murder and other serious crimes.

But a…

Questions arise over the use of an AI crime-fighting tool
theweek.com · 2024

With the rise of artificial intelligence technologies, law enforcement has increasingly been using AI tools to assist in solving crimes — and even putting people behind bars. But now one of these tools, a software program known as Cyberchec…

This AI Tool Helped Convict People of Murder. Then Someone Took a Closer Look
wired.com · 2024

Just after 9 pm on an August night in 2020, Kimberly Thompson and Brian James pulled the car into a driveway in Akron, Ohio, and stepped out into a barrage of gunfire. They were shot in the legs, rushed to a hospital, and survived. But Thom…

Variants

A "variant" is an incident that shares the same causative factors, produces similar harms, and involves the same intelligent systems as a known AI incident. Rather than index variants as entirely separate incidents, we list variations of incidents under the first similar incident submitted to the database. Unlike other submission types to the incident database, variants are not required to have reporting in evidence external to the Incident Database. Learn more from the research paper.

Similar Incidents

By textual similarity

Did our AI mess up? Flag the unrelated incidents

Uber AV Killed Pedestrian in Arizona

Uber AV Killed Pedestrian in Arizona

Mar 2018 · 25 reports
Security Robot Rolls Over Child in Mall

Security Robot Rolls Over Child in Mall

Jul 2016 · 27 reports
Detroit Police Wrongfully Arrested Black Man Due To Faulty FRT

Detroit Police Wrongfully Arrested Black Man Due To Faulty FRT

Jan 2020 · 11 reports
Previous IncidentNext Incident

Similar Incidents

By textual similarity

Did our AI mess up? Flag the unrelated incidents

Uber AV Killed Pedestrian in Arizona

Uber AV Killed Pedestrian in Arizona

Mar 2018 · 25 reports
Security Robot Rolls Over Child in Mall

Security Robot Rolls Over Child in Mall

Jul 2016 · 27 reports
Detroit Police Wrongfully Arrested Black Man Due To Faulty FRT

Detroit Police Wrongfully Arrested Black Man Due To Faulty FRT

Jan 2020 · 11 reports

Research

  • Defining an “AI Incident”
  • Defining an “AI Incident Response”
  • Database Roadmap
  • Related Work
  • Download Complete Database

Project and Community

  • About
  • Contact and Follow
  • Apps and Summaries
  • Editor’s Guide

Incidents

  • All Incidents in List Form
  • Flagged Incidents
  • Submission Queue
  • Classifications View
  • Taxonomies

2023 - AI Incident Database

  • Terms of use
  • Privacy Policy
  • Open twitterOpen githubOpen rssOpen facebookOpen linkedin
  • 8b8f151