Cybercheck Tool Allegedly Provides Questionable Evidence in Murder Trials

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.

Risk classification

• Primary risk domain: 7 AI system safety, failures, & limitations
• Primary risk subdomain: 7.3 Lack of capability or robustness

The primary issue is the AI system's lack of capability and robustness, as it generated inaccurate, unverified, and inconsistent geolocation reports used in critical legal proceedings.

Additional risk subdomains

• 7.4 Lack of transparency or interpretability: The software's proprietary code and opaque methodology prevent independent verification of its findings by defense experts.
• 5.1 Overreliance and unsafe use: Law enforcement and prosecutors relied on unverified AI reports to charge and convict individuals in critical criminal cases.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The safety risks and unreliability stem from the Cybercheck AI system's automated data processing and geolocation algorithms, which were deployed and used in active criminal investigations.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The report describes an AI system used in law enforcement and justice administration ('law enforcement has increasingly been using AI tools to assist in solving crimes — and even putting people behind bars'), which falls under the High Risk category of the EU AI Act due to its significant implications for safety, fundamental rights, and public interests.

AI system and alleged parties

• AI system: Cybercheck (Global Intelligence Inc.)
• AI purpose: Identification; Smart Surveillance
• Behaviour type: Tool
• Alleged developer: Global Intelligence, Cybercheck
• Alleged deployer: Global Intelligence, Cybercheck
• Alleged harmed parties: Yakima County Sheriff’s Office, Unnamed Boulder County Colorado resident, Unnamed Aurora Colorado residents, Sergio Cerna, Phillip Mendoza, Ohio Bureau of Criminal Investigation, Mississippi Bureau of Investigation, Four unnamed Summit County Ohio men

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Minor, indirect Minor
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Minor, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Minor
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: Yes, the reports describe law enforcement agencies paying Global Intelligence between $11,000 and $35,000 for Cybercheck contracts (e.g., Akron paid $25,000, Ohio BCI paid $30,000, Yakima County paid $11,000).

Directly caused: Wasted public funds on unverified software contracts, averaging $66,000 for the 1 systemic occurrence described.

Indirectly caused: Legal defense costs for defendants challenging the unverified evidence, and prosecution costs for cases that had to be dismissed or appealed.

Inferred additional harm: Significant additional financial losses are inferred from the other 300 agencies and 8,000 cases, likely totaling millions of dollars in wasted public funds and legal fees.

Civil rights

Reported: Yes, the reports explicitly describe violations of due process rights.

Directly caused: The use of unverified, secret AI algorithms to generate evidence violates defendants' constitutional right to due process and a fair trial, directly affecting individuals like Adarus Black who was sentenced to life in prison.

Inferred additional harm: It is highly likely that hundreds of other defendants among the 8,000 cases had their due process rights compromised by the introduction of unverified proprietary AI evidence.

Epistemic

Reported: Yes, the reports describe the AI system generating false or unverified geolocation data and identical reports for different dates.

Directly caused: The AI tool generated fabricated or highly inaccurate digital forensic evidence (e.g., placing Mendoza at a crime scene on a date that was 18 days after the crime occurred, and linking non-existent emails).

Indirectly caused: The introduction of false AI-generated evidence into the judicial system undermines the integrity of court proceedings and the shared truth required for fair trials.

Inferred additional harm: It is likely that many other reports generated across the 8,000 cases contained similar unverified or fabricated data points that went unchallenged.

People affected

• Occurrences reported: 1
• People reportedly harmed: 7
• People reportedly exposed: 8000

Potential causes

Management

• Prioritization of sales over truth: Company aggressively marketed unverified tech to police departments.
• Lack of corporate transparency: Global Intelligence refused to disclose methodology or address criticisms.
• Inadequate risk assessment: Prosecutors failed to assess the legal risks of using unvetted AI.

Technology

• Opaque Machine Learning Algorithms: Proprietary algorithms cannot be inspected or verified by defense experts.
• Lack of Data Preservation: Cybercheck does not store raw crawl data, preventing independent replication.
• Unverifiable Geolocation Claims: Claims of Wi-Fi pings from open source data are technically implausible.

Data Inputs

• Inaccurate input dates: System generated identical reports for different dates using same inputs.
• Unreliable open source data: Profiles built on unverified emails and social media accounts.
• Unverifiable Device Profiles: Defense experts could not locate the social media accounts used in profiles.

Human Factors

• Expert witness misrepresentation: Founder lied under oath about peer reviews and prior testimonies.
• Overreliance by law enforcement: Investigators accepted high accuracy claims without seeking supporting evidence.
• Lack of technical expertise: Judges and prosecutors lack computer science skills to evaluate algorithms.

Process and Methods

• Absence of peer review: The software was never independently peer-reviewed despite claims.
• Lack of independent vetting: No third-party validation of the system's accuracy and methodology.
• Flawed validation processes: Law enforcement failed to backstop or verify the software's reports.

Regulatory Environment

• Admissibility standards loophole: Courts allowed unproven AI evidence before Daubert challenges occurred.
• Proprietary trade secret defense: Companies shield algorithms from disclosure by claiming trade secrets.
• Lack of AI forensic standards: No regulatory standards exist to govern AI use in criminal forensics.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide detailed, consistent accounts from multiple reputable investigative journalism sources (The Week, NBC News, WIRED) detailing court filings, expert opinions, and specific case names. The limitations and failures of the AI tool are well-documented through legal motions and statements from prosecutors and defense attorneys.
• Ambiguities identified: The exact inner workings of Cybercheck's proprietary algorithm remain undisclosed, making it impossible to know precisely how it generates its accuracy percentages or geolocation claims.
• Alternative interpretations: One could argue the primary failure is human fraud by the founder rather than a technical AI failure, but the reports clearly show the AI tool itself produced faulty, unverified outputs (e.g., identical reports for different dates).

The widespread deployment of Cybercheck, an unverified AI geolocation tool, in nearly 8,000 US criminal cases has compromised judicial integrity and defendant due process. While causing significant legal disruptions and civil rights concerns, the incident remains a domestic justice system issue manageable through standard judicial procedures, posing minor overall national security risks.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Global Intelligence Inc.

Threat characteristics

• Imminence: Long-term. The situation represents an ongoing strategic concern regarding AI adoption in law enforcement rather than an immediate crisis.
• Autonomy: Human-controlled. The AI tool acts as an investigative assistant generating reports, while humans make all final legal and operational decisions.
• Novelty: Evolved capability. Represents an evolved threat where proprietary, unverified AI geolocation models are integrated into critical criminal justice processes at scale.

Impact by dimension

• Physical security: Negligible. No physical security or critical infrastructure threats were identified in connection with the incident.
• Information security: Negligible. The incident involves domestic law enforcement tools and does not compromise national intelligence capabilities or involve foreign information warfare.
• Sovereignty: Minor. The tool compromised domestic judicial processes across multiple states, but does not threaten overall state authority or federal government continuity.
• Economic security: Negligible. Financial impact is limited to wasted municipal police contracts and does not threaten strategic economic or technological security.
• Societal stability: Minor. The use of unverified AI evidence in 8,000 cases compromises individual due process and civil liberties, but is currently being managed within standard judicial review processes.