Law Vs AI: The US Supreme Court may decide whether AI searches abide by the law

The US Supreme Court is ready to revisit Section 230, a provision that has been at the heart of the Internet for decades. Whatever the court determines may wind up rewriting the regulations for a new technology: artificial intelligence-powered search engines like Google Search and Microsoft’s new Bing.

The Supreme Court will hear arguments in Gonzalez v. Google, one of two related legal concerns, next week. Gonzalez is primarily concerned with whether YouTube may be sued for hosting accounts belonging to foreign terrorists. 

How does Gonzalez v. Google affect AI searches?
The far larger underlying question is whether algorithmic recommendations should be afforded complete legal protection under Section 230 because YouTube suggested such accounts to others. 

Also read: Why a new lawsuit targeting Google & YouTube can potentially change the internet forever

While everyone from tech titans to Wikipedia editors have warned of the possible consequences if the court weakens these safeguards, it raises particularly fascinating problems for AI search, a subject with practically no direct legal precedence to draw upon.

Startups like OpenAI’s ChatGPT are pushing massive language models as the future of search, claiming they can replace increasingly congested traditional search engines. They often replace a list of links with a footnote-heavy summary of material from throughout the web, resulting in conversational responses to inquiries.

These summaries frequently equivocate or state that they rely on the opinions of others. They can, however, introduce inaccuracies and be misleading: in its first show, Bard got an astronomical fact wrong, while in its second demo, Bing cooked up wholly phoney financial statistics for a publicly listed firm.

Even if they are only summarising other stuff from the web, the web is rife with incorrect information. That implies they’re likely to send some of it on, just like traditional search engines. If the errors result in the dissemination of libellous content or other unlawful speech, the search providers may face legal action.

The conundrum
If Section 230 continues largely unaltered, many possible future lawsuits will turn on whether an AI search engine was duplicating someone else’s illegal speech or creating its own. Even if they just slightly alter the phrasing of a user’s original material, web services can claim Section 230 protections. 

Thus an AI programme that only changes a few words may not be held accountable for what it says. Microsoft CEO Satya Nadella has hinted that AI-powered Bing suffers the same legal difficulties as vanilla Bing — and right now, the most pressing legal challenges for AI-generated material are around copyright infringement, which falls outside the jurisdiction of Section 230.

Also read: Microsoft invests more than $10 billion in ChatGPT’s parent organisation OpenAI

There are still constraints here. Linguistic models, like Google’s and Bing’s faults above, can “hallucinate” inaccurate facts, and if these engines create an error, they’re on precarious legal ground under any version of Section 230. How wobbly is it? We won’t know until it comes up in court.

The case may make AI search dangerous even if machines are only providing an accurate paraphrase of someone else’s viewpoint. The central question in the case is whether an online service can lose Section 230 protections by arranging user-generated information in a way that promotes or showcases it. 

Anti-tech sentiment to hurt proceedings?
Judges may be hesitant to apply this to widespread services like old-school search engines, and Gonzalez’s plaintiffs have attempted to demonstrate that this will not happen. Even if they are careful, they may be less inclined to give newer services any leeway since they will become commonplace under the new precedent — particularly services like AI search engines, which disguise search results as direct speech from a digital character.

Also read: Google loses appeal against record EU antitrust fine, to pay 4.125 billion Euros

People gently revising postings by hand will only provide a limited guidepost for complex, large-scale AI-generated content. Judges may be forced to decide how much summarising is excessive under Section 230, and their conclusion may be influenced by the political and cultural atmosphere, rather than merely the wording of the statute. 

Courts have previously interpreted Section 230’s safeguards broadly, but in the face of an anti-tech reaction and a Supreme Court reevaluation of the statute, they may not provide any new technology with the same freedom that previous platforms received.

AI search may not be doomed, but it has issues
All of this is not to say that all AI search is legally doomed. Section 230 is an extremely essential statute, but repealing it would not allow people to win lawsuits for every improperly stated fact. Defamation, for example, necessitates proving that incorrect information exists and that you were hurt as a result of it, among other circumstances.

The distinction between AI summaries and traditional search is not always evident. The standard Google search results page already includes recommended response boxes that editorialise the search results. In the past, they have produced potentially deadly misinformation: for example, Google accidentally transformed a list of “don’ts” for coping with a seizure into a list of suggestions in one search snippet. So yet, there hasn’t been a flood of lawsuits.

Read all the Latest News, Trending News, Cricket News, Bollywood News,
India News and Entertainment News here. Follow us on Facebook, Twitter and Instagram.