When you buy through affiliate links in our content, we may earn a commission at no extra cost to you. Learn 
Perplexity, the AI startup supported by Jeff Bezos, claims to provide answers by searching the web in real-time and attributing sources. However, it has recently gained a nasty reputation.
Forbes accused Perplexity of stealing its content, and a Wired investigation concluded that the AI is breaking the widely accepted Robots Exclusion Protocol to scrape data from web areas where bots should not be operating.
On June 28, Amazon announced that it had opened an investigation and is analyzing the Wired investigation information that claimed Perplexity was using Amazon Web Services (AWS) servers to run data scraping bot operations.
While in the European Union using bots to scrape information from public sites must be justified and meet GDRP standards, in the U.S. scraping data is legal except if the activity breaches laws such as the Computer Fraud and Abuse Act (CFAA), or the Digital Millennium Copyright Act (DMCA).
But legality is not really the big problem with scraping; enforcement is. Monitoring countless bots online and what they are doing is a monumental task nearly impossible to achieve.
Key Takeaways
- Bypassing website restrictions and potentially stealing content undermines trust in AI.
- Perplexity’s inaccurate claims highlight the risk of AI prioritizing “convincing” responses over factual accuracy.
- Regulations like the EU’s AI Act can promote transparency, ethical data use, and reliable AI.
- Responsible development that prioritizes ethics and accuracy is crucial for AI to gain widespread acceptance.
Perplexity is the latest of numerous legal and ethical issues that artificial intelligence has managed to stir up due to copyright infringements, use of data without consent, privacy challenges, and data leaks. From music industry leaders to top media outlets, and more, legal cases against OpenAI, NVIDIA, Google, Microsoft, Anthropic, MidJourney, Stability AI, and DeviantArt are piling up.
However, there is a difference between most cases presented against generative AI technology makers and the Perplexity scandal. Perplexity scraps data in real-time to provide answers, while most AI uses public data for training. Additionally, Perplexity has a big problem with accuracy. The answers the AI has given may make it vulnerable to claims of defamation.
When the Views of AI Power Houses Generate No Trust
Users, companies, and organizations around the world are beginning to reach their breaking point on what an AI should be allowed to do. AI companies continue to push and test the limits, often with irresponsible, unethical, and rather convenient actions and narratives that serve their agenda.
Comments like those recently made by Microsoft’s AI CEO Mustafa Suleyman, in an interview with CNBC, reveal where AI leaders stand on the issue of data use. Their position is in fact concerning.
Suleyman told CNBC that the economics of information are about to radically change and the cost of making content is going to be zero. Suleyman also said that everything on the web is “free” and can be used by AI technology.
“I think that with respect to content that’s already on the open web, the social contract of that content since the ‘90s has been that it is fair use. Anyone can copy it, recreate with it, [or] reproduce with it. That has been ‘freeware,’ if you like, that’s been the understanding.”
Nothing could be further from the truth than what Suleyman said. As mentioned, numerous laws disagree with the “free-internet-content” idea. The idea that AI can freely use any online content reflects the vision that AI leaders want to push forward. This vision often prioritizes profits and risky, illegal, or unethical innovation, over people’s privacy and copyright rights.
The gaps between the regulation landscape in the U.S. and Europe shows that AI innovation can be slowed down with laws. Companies that know they have no possibility of meeting the robust standards of the E.U. AI Act, are choosing not to launch their new AI services in the region.
For example, Apple announced that their AI service — Apple Intelligence — will not be available in Europe any time soon.
Perplexity Takes Accuracy Issues To Defamation Levels
The Wired investigation that triggered the Amazon probe into Perplexity claimed that the AI had generated numerous false claims. In one case, the AI falsely claimed that Wired reported that a specific California police officer had committed a crime.
Chris Rodgers, CEO and Founder at CSP, an advanced SEO revenue-driven solution provider, spoke to Techopedia about the core fundamentals that drive inaccuracies in AI technologies.
“The way LLMs [large language models] function at a fundamental level is troubling when considering they are being relied upon to produce ‘accurate’ answers to potentially important questions.
“Even when supported by an outside knowledge base like the internet through a RAG [retrieval augmented generation] process, LLMs do not function to determine truth or accuracy, this language technology centers around creating convincing human-like responses.”
Rodgers explained that if the RAG knowledge base could provide LLMs with a single source of truth, AI systems could, in theory, produce reliable and accurate results.
“But no such single source of truth exists in our society today, we can’t even agree on what is true.”
Rodgers said that the issues with AI, scrapping, use of data ethics, and legal cases boil down to trust and accuracy.
“We are entering a new era where AI will become more and more integrated into our lives and we will need to decide if we can trust it.”
Naturally, it is easier to trust an AI company if its actions are transparent, ethical, and professional. However, as Rodgers explained when Perplexity accesses websites through bot crawlers — and scrapes information from online areas coded to block crawlers, they are certainly breaking that trust with web publishers.
“It’s akin to providing a homeowner a way to lock out a contractor but then giving that contractor a key to the back-door which they later use to enter without permission.
“Perplexity needs to be more transparent and get real buy-in from users and content creators, people don’t like being tricked, even when you have a shiny new widget.”
The Bottom Line
Perplexity, the AI darling backed by Jeff Bezos, is becoming a poster child for the pitfalls of unchecked AI. While its ability to search the web and attribute sources seems impressive, its methods raise serious concerns about trust and accuracy.
Perplexity’s biggest transgression lies in its data scraping practices. Accusations of bypassing website restrictions and potentially stealing content erode user trust. This highlights a crucial issue: legality doesn’t equal ethicality. Just because scraping might be legal in some regions, it doesn’t make it right, especially when it disregards website owners’ wishes.
Furthermore, Perplexity struggles with accuracy. Reports of demonstrably false claims raise serious doubts about its reliability. As Chris Rodgers, CEO of CSP, points out, AI systems like Perplexity prioritize generating convincing responses over truth-seeking. In a world with no single source of truth, this reliance on potentially biased data sources can lead to dangerous misinformation.
Perplexity serves as a cautionary tale. Without trust and verifiable accuracy, AI will struggle to gain widespread acceptance. The path forward lies in responsible development, prioritizing compliance, ethics, and transparency alongside innovation.
