Flock Audit Assistance: The New AI Tool to Catch Police Abuse — And Why Critics Aren't Satisfied

How Audit Assistance Claims to Work

According to Flock, Audit Assistance uses artificial intelligence to identify search patterns that fall outside a department's policy. Examples include:

• An officer repeatedly searching the same license plate
• Conducting searches outside of normal duty hours
• Running queries without an associated case number

When the system flags such behavior, it can automatically lock out the flagged user until an agency administrator reviews and intervenes . Flock CEO Garrett Langley said the tool has already been "associated with arrests of several law enforcement officers who allegedly abused the system" .

What Flock Has Not Disclosed

Despite multiple media inquiries, Flock has refused to share key technical details about Audit Assistance . The company has not disclosed:

• Algorithm specifics: What exact behavioral signals or thresholds the AI uses to determine what counts as "abnormal"
• Training data: What data was used to train the detection model. Flock's privacy policy states that less than 1% of captured images (stripped of metadata) may be used internally for machine learning, but it has not confirmed whether that applies to Audit Assistance
• Effectiveness metrics: No false-positive rates, detection rates, or any other performance benchmarks have been released. As TechCrunch noted, Flock "hasn't explained how it works" and has provided no evidence the tool is effective at catching abuse

Privacy Experts and Critics Respond

The ACLU, EFF, and Brennan Center have been sharply critical of the reforms. The ACLU called the entire package a "thinly veiled PR attempt" and stated bluntly that "the effectiveness of its audit tool has never been independently evaluated" .

Broader concerns include:

• Audit Assistance still relies entirely on individual agencies to review flags and take action. Flock does not block misuse itself — it merely flags it and locks accounts pending human review
• There is no independent auditing mechanism or third-party oversight built into the system

Broader Reforms and Context

Audit Assistance is just one of several changes Flock announced on August 13, 2026 :

• Reduced data retention: Default retention for ALPR data was cut from 30 days to 7 days
• Evidence Mode: A new feature allows data to be preserved only for active criminal cases
• Case number requirement: Officers will now be required to label every search with a criminal case number

These changes came after a wave of documented police abuse, including:

• Three Bibb County, Georgia deputies — Coznavian Stubbs, Toni Cierra Lewis, and Joseph Marquee Callaway — arrested on August 10, 2026 after using Flock cameras to stalk people they knew for personal reasons. They were charged with stalking and computer invasion of privacy .
• Richmond County, Georgia: Three sheriff's office employees (Lt. John McKenzie, Investigator Curtis Hoskins, and Deputy Joshua Johnson) were arrested and fired in July 2026 for similarly misusing Flock cameras .
• Ogeechee Judicial Circuit, Georgia: Investigator Andrae Wright was arrested for accessing the Flock system more than 60 times in three weeks to stalk someone .

Flock CEO Garrett Langley acknowledged in an interview with the BBC that the company "took too long to act" and admitted that the new safeguards should have been introduced sooner .

The Bottom Line

Flock's Audit Assistance tool represents a significant step toward automated oversight of police surveillance activity. However, without independent evaluation of its algorithm, training data, and effectiveness, critics remain skeptical that it is anything more than a public relations response to mounting abuse scandals.