Free tools Windows power users keep installed
One-click scans. No signup required.
A July 2024 survey found that 84% of respondents wanted the U.S. government to update copyright laws to protect against AI. But the figure is narrower than the headline “tech execs” suggests: The Harris Poll surveyed 307 U.S. corporate data-management, privacy and AI decision-makers at director level or above for sponsor Collibra. It measured support for a broad goal—not agreement on a specific bill or licensing plan.
What the survey found
The Harris Poll conducted the online U.S. survey from July 9–12, 2024, for Collibra, a data-intelligence and AI-governance company. The results point to broad appetite for oversight among this specialized group:
![AI Law and Policy: [Connected eBook]](https://m.media-amazon.com/images/I/41Y7jOay2eL._SL160_.jpg)
| Survey finding | Share of respondents |
|---|---|
| Wanted copyright laws updated to protect against AI | 84% |
| Supported Big Tech compensating people for use of their data in AI training | 81% |
| Supported federal AI regulation | 76% |
| Supported state-level AI regulation | 75% |
| Said AI-related threats necessitate U.S. government regulation | 99% |
The 64% figures for privacy and security as major concerns, and the finding that 75% said their companies prioritize AI training and upskilling, were reported by VentureBeat. These percentages answer different questions; they should not be read as a single measure of support for one policy.
Who answered—and what “tech executives” means here
The sample was 307 U.S. adults aged 21 or older who worked full time and were responsible for data management, privacy and/or AI decisions at their companies, at director level or higher. That makes “corporate data, privacy and AI decision-makers” a more accurate description than “tech executives” generally. The respondents were not a sample of all U.S. executives, all technology workers or the public.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitches#1 Best Overall
The survey’s sponsor was Collibra, which sells data-intelligence and AI-governance products. The Harris Poll conducted the survey, but the commercial sponsorship is relevant context: the poll’s wider focus on governance and regulation overlaps with Collibra’s business. Sponsorship does not by itself invalidate the findings; it does mean the results should not be presented as an independent census of the technology sector. Collibra CEO Felix Van de Maele argued that creators deserve transparency, protection and compensation while describing data as foundational to AI performance. That is the sponsor’s perspective, not a neutral legal finding.
What respondents meant by updating copyright law
The reported question asked whether the U.S. government should update copyright laws “to protect against AI.” That wording does not reveal what kind of change respondents had in mind. It does not show majority support for compulsory licensing, per-use royalties, opt-in training, an opt-out registry, required training-data disclosures, collective licensing, new remedies for unauthorized outputs, or a new copyright in prompts or AI-generated works.
Rank #2
“Protect against AI” also combines distinct policy problems. A law addressing the use of works in training may not settle whether a generated output infringes a particular work, how creators should be paid, or what information model developers must disclose. The survey did not rank these questions or show how respondents would resolve conflicts among them.
The separate copyright questions behind the headline
Training data: permission, access and transparency
One issue is whether copyrighted works may be collected and used to train models, and whether commercial training should be treated differently from noncommercial research. A policy could turn on permission, lawful access, or another standard. It could also require developers to identify sources or categories of training data. The poll does not establish which standard respondents prefer.
Recommended Free Tools
Outputs: copying and responsibility
Another question is whether generated text, images, audio, video or code substantially reproduces protected expression. There is also a question of responsibility: a dispute might involve a model developer, a business deploying a third-party model, or an end user. The survey does not establish whether respondents believe existing copyright doctrines are sufficient or favor new rules.
Compensation: who pays, and for what?
Possible approaches include direct agreements with creators or publishers, collective licensing, statutory payments, or compensation tied to training, outputs or commercial revenue. Each raises practical questions: who qualifies as a contributor, how a work’s value is measured, how licensed and public-domain material is separated from other material, and who administers payments. The 81% support figure concerns compensation by Big Tech for people whose data is used in AI training; it does not identify a payment mechanism or settle how copyright applies to any particular work.
Rank #4
Transparency and enforcement
Recordkeeping, notices to creators, auditability, provenance information and remedies for unauthorized copying could help with transparency or enforcement. They can also create trade-offs: detailed disclosures may help rights holders understand how works were used, while companies may argue that dataset and filtering disclosures reveal proprietary information. The survey did not test these options individually.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why companies and creators want clearer rules
Creators and content owners seek control, attribution, permission and compensation. AI developers need access to large, high-quality datasets. Businesses using third-party models may also want to know whether training data, model behavior or an output could create legal or compliance exposure. Uncertainty can complicate decisions about whether to license content, build a model, use a commercial model or change internal controls.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Those interests can pull in different directions. A permission-first or compensation-first system could give creators more control and revenue, but add cost and complexity for developers, particularly smaller ones and researchers. A clear statutory framework could reduce uncertainty, while rules tied too tightly to today’s technical practices could be difficult to adapt as training methods change. A national rule could reduce fragmented compliance, but state rules may still matter in areas such as privacy, consumer protection, employment or synthetic media.
How much weight should readers give the 84% figure?
The result is a strong signal about the surveyed group, not a referendum of the U.S. technology sector. Harris reported sampling precision of approximately ±5.7 percentage points at a 95% confidence level. That is a reported estimate, not a guarantee that another survey—especially a probability sample—would produce the same result. The survey was conducted online, and the published Collibra release says full methodology, including weighting variables and subgroup sample sizes, was available by contacting the company rather than listing all those details on the page.
The number also describes stated opinion, not corporate behavior: it does not show whether respondents’ employers changed their data practices, licensed works or supported a particular legislative proposal. And it dates from July 2024. It should not be treated as a measure of executive opinion in 2026.
For a finance or business reader, the practical takeaway is about uncertainty and competing costs, not a settled compliance rule. A company may need to track data provenance, model use, approvals and potential exposure, but governance systems cannot determine on their own whether a training use is lawful, an output infringes, or a creator is owed compensation.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
