Can Lawyers Use AI Video Enhancement
Without Violating Client Confidentiality?
Video evidence is increasingly central to litigation, depositions, and client consultations. AI enhancement can dramatically improve the clarity and usefulness of that footage. But every time a lawyer uploads client video to a third-party tool, they face a confidentiality question that most legal technology discussions skip entirely.
No AI training. No data sharing. 30-day auto-deletion. Audit-ready architecture.
The Problem Every Litigator Faces
A client brings you surveillance footage from a parking lot incident. The video is grainy, poorly lit, and shot at 720p on a years-old camera system. You know the footage shows something important — maybe the license plate of a vehicle, maybe the face of a person near the scene — but the detail is too blurry to be conclusive. A colleague suggests running it through an AI enhancement tool. The footage comes back sharper, clearer, and suddenly usable.
But in the thirty seconds it took to upload that footage to an AI tool, what did you actually do? You transmitted a piece of confidential client evidence to a third-party server. You agreed — by clicking through a terms of service screen you almost certainly didn't read — to the platform's data policies. You may have given that platform a license to use your client's footage. You may have introduced a chain of custody issue that opposing counsel will exploit the moment that footage becomes an exhibit.
This is not a hypothetical risk. It is a real and underexamined one, and as AI video tools become more capable and more widely used in legal practice, it will become a more frequent source of ethics complaints, evidentiary challenges, and malpractice exposure.
What Client Confidentiality Actually Requires
Model Rule 1.6 of the ABA Model Rules of Professional Conduct prohibits lawyers from revealing information relating to the representation of a client unless the client gives informed consent, the disclosure is impliedly authorized to carry out the representation, or an exception applies. Most state bar rules track this standard closely.
Importantly, the obligation covers not just communications but all "information relating to the representation" — a broad category that encompasses client video footage, whether it is privileged or not. The question of whether uploading footage to a third-party AI tool constitutes a "revelation" is a factual and legal one that depends on what the vendor does with the data.
There is a long-recognized principle that lawyers may share client information with agents who assist in the representation — paralegals, expert witnesses, co-counsel — without violating confidentiality, provided those agents are adequately supervised and bound by equivalent confidentiality obligations. The question is whether a cloud AI vendor qualifies as such an agent, and under what conditions.
The answer, based on the growing body of bar guidance on legal technology and cloud computing, is: yes, with conditions. Most state bars that have addressed cloud computing for legal use have concluded that lawyers may use cloud services to store and process client data if they have taken reasonable steps to ensure confidentiality, including evaluating the vendor's security practices, entering into appropriate agreements, and ensuring the vendor does not use client data for unauthorized purposes.
How AI Video Processing Works — and Where the Risks Are
When you upload a video to an AI enhancement tool, the video is transmitted over the internet to the vendor's servers. There, it is processed by machine learning models — typically deep neural networks trained on large datasets of video pairs. The processing may take anywhere from seconds to minutes depending on the length of the video and the type of enhancement being applied. The enhanced output is then transmitted back to you.
The risks in this pipeline are located at several points. First, during transmission: unencrypted or poorly encrypted transmission can expose the video to interception. Second, at storage: while the video is on the vendor's servers, it is subject to the vendor's access controls, backup policies, and breach exposure. Third, in the data pipeline: many AI platforms use processed data — including the inputs — to retrain or improve their models. If your client's confidential footage is used to train an AI model, it may effectively become embedded in a system used by thousands of other parties.
Fourth, at deletion: "deleting" data in a cloud system is not always what it appears. Backups, disaster recovery snapshots, and training data archives may retain copies of data long after the user's apparent deletion. Fifth, at third-party sharing: most consumer AI platforms use multiple third-party cloud providers, CDN networks, and analytics tools. Your client's data may touch many more systems than the one you agreed to use.
The Key Confidentiality Risks for Attorneys
- Unauthorized use of client data: Many AI platforms assert in their terms of service the right to use uploaded content to improve their services. If "improve their services" includes model training, your client's footage may be used in ways they never consented to.
- Indefinite retention: Consumer platforms often retain data indefinitely or for very long periods. A video uploaded for enhancement in 2024 might still be on the vendor's servers in 2030 — creating a long window of potential breach exposure.
- No contractual confidentiality obligation: Without a DPA or vendor agreement, the platform's obligations to you are defined by its standard terms — which are drafted to protect the platform, not your client.
- Third-party subprocessors: Most cloud platforms use multiple subprocessors. Each one receives your client's data, and each one represents an additional link in the chain of potential breach exposure.
- Privilege challenges: If the footage is introduced as evidence, opposing counsel may challenge its provenance and integrity based on the fact that it was transmitted to and processed by a third party.
Best Practices for Attorneys Using AI Video Tools
- Review the platform's terms of service and privacy policy before uploading any client footage — focus on data use, training, retention, and subprocessor disclosure
- Require a signed Data Processing Agreement that explicitly prohibits use of client data for model training or any purpose beyond the processing service
- Confirm the vendor's deletion policy in writing — auto-deletion timeline, backup policies, and what "deleted" means technically
- Verify that the vendor encrypts data in transit (TLS) and at rest (AES-256 or equivalent)
- Document the enhancement process: date, tool used, version, what enhancement was applied, and the original file hash to establish continuity
- Preserve the original unenhanced footage separately and independently of the enhanced version
- Obtain informed client consent where required by your jurisdiction's ethics rules, especially for especially sensitive footage
- Disclose AI enhancement to opposing counsel if the footage will be introduced as evidence, consistent with applicable rules of evidence
- Review any applicable jurisdiction-specific bar guidance on AI tools in legal practice before deployment
How BetterVideo Addresses These Concerns
BetterVideo was built with a simple principle: client footage is processed for enhancement and nothing else. Our technical architecture reflects this — your video is uploaded to isolated cloud storage, processed by an ephemeral GPU container that is destroyed after the job, and then stored in your private vault until you delete it or the 30-day automatic expiry triggers. At no point is your footage used to train, update, or fine-tune our AI models. At no point is it shared with third parties for any purpose.
For attorneys, this means the key confidentiality risk factors are addressed architecturally, not just by policy. There is no model training because the processing infrastructure is physically separated from the model training environment. There is no third-party sharing because the processing pipeline does not involve third-party analytics, advertising, or data brokerage systems. The 30-day deletion is automated and enforced by the database schema, not dependent on a manual process.
We are not offering a legal opinion on whether our platform satisfies your jurisdiction's ethics rules for any specific use case. The answer to that question depends on the nature of the footage, the applicable rules, and the specific facts of your representation. But we are offering a platform whose architecture is designed to minimize the confidentiality risks that attorneys face when using AI video tools — and to be transparent about that architecture so it can be evaluated by counsel.
Risk Mitigation Checklist for Attorneys
- ☐ Reviewed vendor ToS and privacy policy for data use, training, and sharing clauses
- ☐ Confirmed vendor does not train AI models on uploaded content
- ☐ Confirmed vendor's data retention and deletion policy in writing
- ☐ Entered into a signed DPA or vendor confidentiality agreement
- ☐ Verified encryption in transit and at rest
- ☐ Documented the enhancement process with file hashes and timestamps
- ☐ Preserved original unenhanced footage separately
- ☐ Considered and addressed client consent requirements for your jurisdiction
- ☐ Planned disclosure to opposing counsel if footage will be used as evidence
- ☐ Reviewed applicable bar guidance on AI tools and cloud processing for legal work
Frequently Asked Questions
Not automatically, but it requires careful evaluation. Uploading footage to a third-party service can constitute a disclosure that waives privilege unless the vendor is operating as a functional agent and has adequate confidentiality protections. The key factors are: a signed confidentiality or data processing agreement, a vendor architecture that actually protects the data, and a disclosure that was reasonably necessary to the representation.
A Data Processing Agreement (DPA) is a contract governing how a vendor handles data you share with them — what they can use it for, how they store and protect it, when they delete it, and what happens in a breach. For legal work involving client video, a DPA is essentially mandatory. Standard consumer terms of service are not adequate substitutes.
Yes, but it requires a vendor with adequate confidentiality safeguards and a clear chain of custody process. The enhanced version should be treated as derivative work product, the original preserved separately, and any AI enhancement disclosed to opposing counsel if the footage becomes an exhibit.
No. BetterVideo does not share, sell, or transfer uploaded video to any third party. Processing occurs in isolated GPU containers. Videos are automatically deleted after 30 days. BetterVideo does not use uploaded footage to train AI models.
At minimum: a signed DPA, clear data retention and deletion policy, no AI training on client data, encryption in transit and at rest, access controls, breach notification procedures, and information about who at the vendor can access client data and under what circumstances.
Built for professionals who can't afford a privacy mistake.
No AI training on uploads. No data sharing. 30-day deletion. Process client footage with confidence.
No subscription required. Pay per use. Credits never expire.