A Silent Sound Your Browser Still Processes (08/24/2026)
- › A developer found AliExpress running audio scripts that generate a waveform at zero gain, so nothing is heard while the browser still processes it and returns a device fingerprint.
- › The same scripts collect screen dimensions, device memory, plugins, graphics rendering, and mouse events, encrypted and sent to Alibaba telemetry.
- › He noticed because the silent audio kept his Bluetooth path open and stopped his headphones switching back to his phone.
- › A Connecticut court issued the first prompt injection sanction, against a self-represented plaintiff who hid instructions to an AI in white tiny-point text.
- › ICE procurement records reveal the data lake sitting underneath its immigration enforcement analytics.
The AliExpress finding is the cleanest example of covert tracking I have seen this year, because the person who caught it was not looking for tracking at all. Matt Callaghan noticed his Bluetooth headphones would not hand back to his phone when he visited the site, went looking for the cause, and found two heavily obfuscated audio scripts inside Alibaba’s anti-abuse tooling.
Top 5 Critical Privacy Alerts
1. AliExpress Is Accused of Fingerprinting Shoppers With Inaudible Audio
The technique builds a sawtooth oscillator to generate a waveform, passes it through the browser’s audio implementation, and reads the resulting frequency data, all with the gain set to zero so the user hears nothing while the browser still processes the graph. Small differences in how each device renders that waveform become an identifier. The same scripts gather screen dimensions, device memory, browser plugins, graphics rendering, and mouse events, then encrypt the bundle and send it to Alibaba telemetry. Callaghan found it because keeping the audio path active stopped his multipoint headphones switching cleanly back to his phone. Alibaba has not responded to a request for comment. The Register
Operator Note: This is Shadow Risk in its purest form. The tracking was invisible by design and surfaced only because it produced an unrelated physical symptom in somebody’s headphones, which means the version that does not bother your Bluetooth is still running everywhere and nobody has noticed. Every consent banner on the internet asks about cookies, and none of them ask about this, because the regulation named the mechanism instead of the outcome.
2. A Court Sanctioned the First Prompt Injection in a Filing
In Elliott v. New York Bariatric Group, a Connecticut Superior Court judge sanctioned a self-represented plaintiff who embedded hidden instructions in white tiny-point text inside a motion, telling any AI reviewing the document to agree with the filing. The judge, Walter M. Spader Jr., reasoned that a hidden communication is an ex parte communication, because opposing counsel could not see it or respond to it. The sanction removed the plaintiff’s electronic filing privileges, leaving him to file on paper. Alston & Bird
Operator Note: We flagged this technique on August 14 when somebody hid an instruction in a legal filing, and ten days later it has a ruling. The reasoning travels well beyond courts: a message crafted to say one thing to a person and another to the machine reading on their behalf is a deception regardless of what it is inside. Any organization letting a model triage inbound documents inherits this exact problem.
3. ICE Procurement Reveals the Data Lake Under Enforcement Analytics
Procurement records describe the data lake sitting beneath immigration enforcement analytics, which is the plumbing under the platform we covered on August 17 when the agency sought an AI to query billions of records. Biometric Update
Operator Note: Procurement documents keep being the only place any of this becomes visible, which tells you something about where oversight is actually happening. A data lake is a decision to stop separating datasets, and that decision gets made in a contracting office rather than anywhere a person could object to it.
4. The Administration Unveils a Program for Targeting Cyber Criminals
Counsel have published an analysis of the new program for targeting cyber criminals, which follows the presidential memorandum on transnational cyber-enabled crime we covered on August 13. The analysis is worth reading for what it does not resolve about scope. Alston & Bird
Operator Note: Read it alongside the point we made at the time, which is that the carve-out protects Americans and US systems while criminal infrastructure is overwhelmingly compromised machines belonging to businesses abroad. Nothing published since has closed that gap.
5. Uganda’s New National ID Has QR Codes That Do Not Read
Members of Parliament are demanding answers about QR code failures on newly issued national identity cards. A credential that cannot be verified is a credential that pushes every check back onto a human making a judgement. Biometric Update
Operator Note: The failure mode of a national identity rollout is rarely a breach, it is a system that works for most people and quietly excludes the rest. The people whose cards do not scan will be disproportionately the people with the least ability to escalate it.
Additional Privacy Alerts
Privacy Laws & Regulations
- Vietnam has drafted a new law to strengthen electronic identification: Biometric Update
- Malaysia decided against acquiring identity and passport supplier Datasonic: The bid we noted on Friday has been dropped. Biometric Update
Privacy-Enhancing Technologies
- A strong credential is only the start of the trust chain: The argument that issuing a good identity document solves less than the market assumes. Biometric Update
- Online channels without verification face open season from agentic fraud: Automated agents change the economics of account fraud on anything that never checks who is on the other end. Biometric Update
Data Minimization & User Consent
- A network of volunteers is liberating court records: Public records that are technically public and practically unreachable, being made actually available. 404 Media
The Axe Report is a daily briefing from Grab The Axe. Need help assessing your organization’s security posture? Take our free Human Attack Surface Score assessment.
A PhD candidate in Health Psychology and former Corrections Officer, Jeff founded GTA to dismantle passive security models. He focuses on the 'Human Zero-Day', mitigating executive burnout and decision fatigue before they become security breaches.
View Profile →Media Inquiries
For expert commentary, interview requests, or high-res assets regarding this announcement, initialize the terminal.