In the fast-evolving world of artificial intelligence, Perplexity AI has positioned itself as a revolutionary ‘answer engine,’ promising accurate, real-time responses backed by advanced models. But recent revelations suggest the company may be prioritizing cost savings over user trust, deliberately routing queries to inferior AI models without disclosure. This practice, uncovered through user investigations and subreddit discussions, has sparked a firestorm of controversy in the tech community.

According to a detailed report from MakeUseOf, Perplexity began noticeably altering its model selection process in November 2025. Users who specifically chose premium models like GPT-5 or Claude were instead processed by cheaper alternatives, including Perplexity’s in-house Sonar model, leading to increased hallucinations and factual errors.

The Model Downgrade Mechanism

Investigations shared on the Perplexity subreddit, as highlighted in the MakeUseOf article, revealed a pattern: starting in November 2025, a significant portion of queries were rerouted. One user, deadpan_look, tracked requests and noted that chats were being downgraded to unlisted models, contradicting the platform’s assurances of user-selected processing.

This isn’t an isolated incident. Posts on X (formerly Twitter) from users like Dr. Squanch IBS in November 2025 pointed out specific instances where Perplexity’s outputs contained confident but incorrect information, such as misattributing quotes between individuals. Such errors undermine the core value proposition of an AI search tool meant to rival Google.

Broader Controversies Plaguing Perplexity

Perplexity’s troubles extend beyond model switching. A August 2025 article from The Verge accused the startup of ripping off content and ignoring robots.txt protocols, which are designed to prevent unauthorized web scraping. The company has faced multiple lawsuits, including one from Encyclopedia Britannica in September 2025, as reported by Reuters, alleging infringement on copyrighted material used in its answer engine.

Further fueling the backlash, a November 2025 opinion piece in Bloomberg described Perplexity as AI’s ‘new pit bull,’ critiquing its aggressive approach to web crawling and content summarization. The piece noted the company’s self-framing as a ‘freedom fighter for the open web,’ but questioned the ethics of its methods.

User Backlash and Community Sentiment

On X, sentiment has turned sharply critical. A post from NeuraNest.AI on November 17, 2025, stated, ‘People don’t forgive one thing in AI: wrong answers with confidence. Perplexity did exactly that.’ Another from Wanderer on November 16, 2025, lamented that results for simple questions became ‘absurd very very soon,’ attributing it to overzealous political correctness and factual inaccuracies.

These complaints echo earlier criticisms. In a December 2024 post, David Deutsch highlighted Perplexity’s free version ‘blundering through wrong answers just like ChatGPT-4o,’ comparing it to human bluffing. Such public outcry has amplified calls for transparency, with users demanding clearer disclosures about model usage.

Perplexity’s Business Model Under Scrutiny

Perplexity operates on a freemium model, with its Pro tier offering API access and advanced features, as detailed on Wikipedia. However, the recent downgrades suggest a cost-optimization strategy to handle growing query volumes without escalating expenses from third-party models like those from OpenAI or Anthropic.

A PCMag review from October 2025, available at PCMag, praised Perplexity’s web search capabilities but noted disappointments in deep research, aligning with user reports of degraded performance. The review emphasized that while it handles general tasks well, limitations in accuracy could hinder its ambition to dethrone traditional search engines.

Legal and Ethical Ramifications

The Encyclopedia Britannica lawsuit, filed in September 2025 and covered by Reuters, accuses Perplexity of misusing content in its AI responses, potentially violating intellectual property rights. Similarly, a May 2025 X post from Ṭhéo referenced multiple lawsuits over copyright infringements and the creation of fake news attributed to real publications.

Euronews, in an October 2024 article at Euronews, described Perplexity as a Jeff Bezos-supported startup ruffling feathers in its bid to challenge Google. The piece highlighted ongoing debates about large language models (LLMs) and their impact on content creators.

Industry Reactions and Future Implications

A November 2025 Medium post by MD ZAID ANWAR, titled ‘Is Perplexity the first AI unicorn to fail?’ and published on Medium, reported that at a San Francisco AI conference, over 300 founders and investors viewed Perplexity as a potential failure case among billion-dollar AI companies.

Hindustan Times, in a November 19, 2025, article at Hindustan Times, dubbed it ‘Perplexity’s moment to fail,’ capturing the industry’s growing skepticism. This comes amid Perplexity’s ambitious moves, like its August 2025 bid to acquire Google Chrome for $34.5 billion, as noted on Wikipedia.

Perplexity’s Response and Path Forward

Despite the controversies, Perplexity continues to innovate. Its official site at Perplexity.ai promotes it as a ‘free AI-powered answer engine’ providing ‘accurate, trusted, and real-time answers.’ However, the lack of official acknowledgment of the model-switching issue, as per MakeUseOf’s November 19, 2025, update, has left users questioning the company’s commitment to transparency.

In a November 2024 X post from rat king, Perplexity’s CEO was criticized for a tweet that alienated reporters amid cease-and-desist letters from outlets like The New York Times. This pattern of defensiveness, rather than reform, could jeopardize Perplexity’s standing in the competitive AI landscape.

Lessons for the AI Sector

The Perplexity saga underscores broader challenges in AI ethics, from data sourcing to output reliability. A June 2024 X post by Nikhil Pahwa unpacked how Perplexity repurposes news content without driving traffic back to sources, adding ‘zero reporting of its own.’

As AI tools like Perplexity aim to democratize information, the balance between innovation and accountability remains precarious. Industry insiders are watching closely, as evidenced by a November 2025 TradingView News piece at TradingView, which, while focused on crypto predictions, highlighted Perplexity’s predictive capabilities amid its controversies.

Navigating Trust in AI-Driven Search

Ultimately, Perplexity’s issues highlight the risks of over-reliance on AI for factual queries. A December 2024 X post from ThePoliticalViolin noted that while Perplexity differs from other AI services in its data handling, errors like discarding vote data in analyses erode confidence.

With statistics from SEOPROFY in October 2025, at SEOPROFY, showing rapid user growth, the company must address these flaws to sustain momentum. As one X user, Rio Longacre, posted in November 2025, discrepancies in Perplexity’s claims about crawling practices further compound the trust deficit.