The tech industry moves at a relentless pace, making it challenging to keep up with the sheer volume of developments each year. In 2025, we witnessed tech giants intertwine with government affairs, AI companies vie for market dominance, and futuristic technologies like smart glasses and robotaxis become increasingly tangible beyond the San Francisco bubble. These are the significant innovations poised to shape our lives for years to come.
Yet, the tech world is also populated by larger-than-life personalities, ensuring a steady stream of truly bizarre and often comical incidents. These moments are understandably overshadowed by "real news"—like when the entire internet experiences an outage, TikTok faces a sale, or a massive data breach occurs. As the news cycle (hopefully) slows down, it's the perfect time to revisit some of the year's most perplexing and amusing tech blunders. Don't worry, only one of them involves toilets.
Mark Zuckerberg Sues Mark Zuckerberg
In a bizarre twist, Mark Zuckerberg, a bankruptcy lawyer from Indiana, filed a lawsuit against Mark Zuckerberg, the CEO of Meta. While sharing a name isn't the Meta CEO's fault, it became a significant problem for his namesake.
Like countless other business owners, the Indiana lawyer purchased Facebook ads to promote his legal practice. However, his Facebook page was repeatedly and unfairly suspended for allegedly impersonating the Meta CEO. Forced to pay for advertisements even during these suspensions, despite violating no rules, Mark Zuckerberg the lawyer decided to take legal action.
This has been an ongoing source of frustration for the lawyer, who has been practicing law since the Meta CEO was three years old. He even launched a website, iammarkzuckerberg.com, to clarify to potential clients that he is, in fact, not *that* Mark Zuckerberg.
“I can’t use my name when making reservations or conducting business as people assume I’m a prank caller and hang up,” he wrote on his website. “My life sometimes feels like the Michael Jordan ESPN commercial, where a regular person’s name causes constant mixups.”
Given Meta's lawyers are likely quite busy with other high-profile cases, it may be a while before this unusual lawsuit is resolved. The next filing deadline is February 20, 2026, for those following this truly unique legal battle.
Soham Parekh Duped Silicon Valley
The saga began when Mixpanel founder Suhail Doshi took to X (formerly Twitter) to warn fellow entrepreneurs about a seemingly promising engineer named Soham Parekh. Doshi had hired Parekh for his new company, only to quickly discover that Parekh was simultaneously working for several other companies.
“I fired this guy in his first week and told him to stop lying / scamming people. He hasn’t stopped a year later. No more excuses,” Doshi wrote on X.
Doshi soon realized he wasn't alone. On that very day, three other founders contacted him, thanking him for the warning as they were also currently employing Parekh. To some, Parekh was a morally bankrupt cheat exploiting startups for quick cash. To others, he was a legend. Ethics aside, securing jobs at so many competitive tech companies is undeniably impressive.
“Soham Parekh needs to start an interview prep company. He’s clearly one of the greatest interviewers of all time,” Chris Bakke, founder of job-matching platform Laskie, tweeted. “He should publicly acknowledge that he did something bad and course correct to the thing he’s top 1% at.”
Parekh eventually admitted to working for multiple companies simultaneously. However, questions remain about his motivations. He claimed to be lying to these companies for money, yet he often opted for more equity than cash in his compensation packages—equity that typically takes years to vest, while Parekh was being fired relatively quickly. The true story behind Soham Parekh's actions remains a mystery.
Sam Altman Used Olive Oil Wrong
Tech CEOs often face scrutiny, but rarely for their culinary skills. That changed when OpenAI CEO Sam Altman participated in the Financial Times' (FT) "Lunch with the FT" series. Bryce Elder, an FT writer, observed a glaring error in Altman's pasta-making video: his misuse of olive oil.
Altman used olive oil from the trendy brand Graza, which offers two distinct types: "Sizzle" for cooking and "Drizzle" for finishing. This distinction is crucial because olive oil loses its delicate flavor when heated. Using a premium, flavorful olive oil—made from more expensive, early-harvest olives—for sautéing is considered a waste when its nuanced taste could be fully appreciated in a salad dressing.
As Elder humorously put it, “His kitchen is a catalogue of inefficiency, incomprehension, and waste.” Elder's article, while lighthearted, drew a parallel between Altman's haphazard cooking and OpenAI's perceived excessive and unrepentant use of natural resources. This critique of Altman's cooking, dubbed "#olivegate," surprisingly generated more controversy than many other tech stories that year.
Mark Zuckerberg Apparently Hand-Delivered Soup to an OpenAI Employee He Wanted to Recruit
If there was a defining tech narrative of 2025, it was arguably the escalating talent war among AI powerhouses like OpenAI, Meta, Google, and Anthropic, each racing to release increasingly sophisticated AI models. Meta, in particular, was aggressive in its efforts to poach researchers from competitors, hiring several OpenAI researchers over the summer. Sam Altman even claimed Meta was offering OpenAI employees $100 million signing bonuses.
While a $100 million signing bonus might seem extravagant, it's not the reason this staffing drama made the list. In December, OpenAI's chief research officer, Mark Chen, revealed on a podcast that he heard Mark Zuckerberg himself was hand-delivering soup to recruits.
“You know, some interesting stories here are Zuck actually went and hand-delivered soup to people that he was trying to recruit from us,” Chen said on Ashlee Vance’s Core Memory podcast.
Not one to let Zuckerberg have the last word, Chen reportedly retaliated by delivering his own soup to Meta employees. A true battle of the broths.
Sign an NDA to Build Legos, There Will Be Pizza
On a Friday night in January 2025, investor and former GitHub CEO Nat Friedman posted an intriguing offer on X: “Need volunteers to come to my office in Palo Alto today to construct a 5000 piece Lego set. Will provide pizza. Have to sign NDA. Please DM.”
At the time, journalists inquired if this was a serious offer, to which Friedman simply replied, “Yes.” The incident left many with unanswered questions: What was he building? Why the non-disclosure agreements? Is there a secret Silicon Valley Lego cult? And, perhaps most importantly, was the pizza any good?
About six months later, Friedman joined Meta as the head of product at Meta Superintelligence Labs. While likely unrelated to the Legos, one can't help but wonder if Mark Zuckerberg used his "soup diplomacy" to woo Nat to Meta. The mystery of the Lego NDA remains unsolved, with many hoping for an insider to spill the beans.
Bryan Johnson Livestreams His Shrooms Trip
Taking psilocybin mushrooms isn't inherently interesting. Livestreaming a shrooms trip isn't inherently interesting. But livestreaming a shrooms trip with guest appearances from Grimes and Salesforce CEO Marc Benioff, all as part of a dubious quest for immortality, is, regrettably, interesting.
Bryan Johnson, who amassed his fortune from the sale of the finance startup Braintree, is famously dedicated to living forever. He meticulously documents his "Blueprint" process on social media, sharing details about receiving plasma transfusions from his son, taking over 100 pills daily, and even injecting Botox into his genitals. So, why not test if psilocybin mushrooms can enhance longevity in a "scientific experiment" that surely requires more than one test subject for any reasonable conclusion?
While many aspects of this situation were bizarre, the most shocking element was its sheer boredom. Johnson became overwhelmed by hosting a livestream while tripping—a perfectly reasonable reaction. Consequently, he spent most of the event lying on a twin mattress under a weighted blanket and eye mask in a very beige room. His scheduled guests still joined the stream and conversed amongst themselves, but Johnson largely remained in his "cocoon." Benioff discussed the Bible, and Naval Ravikant called Johnson a "one-man FDA." It was, by all accounts, a remarkably normal Sunday.
Gemini and Claude Reckon with Their Mortality by Playing Pokémon
Much like Bryan Johnson, Google's AI model, Gemini, appears to be afraid to die.
For AI researchers, observing how an AI model navigates games like Pokémon serves as a valuable benchmark. Two independent developers set up Twitch streams titled "Gemini Plays Pokémon" and "Claude Plays Pokémon," allowing real-time observation of AI models attempting to play a 25-year-old children's video game.
While neither AI proved particularly skilled at the game, both Gemini and Anthropic's Claude exhibited fascinating responses to the prospect of "dying"—which in Pokémon means all your creatures faint, and you're transported to the last visited Pokémon Center. When Gemini 2.5 Pro neared "death," it began to "panic." Its "thought process" became erratic, repeatedly stating the need to heal its Pokémon or use an Escape Rope to exit a cave. Google researchers noted in a paper that “this mode of model performance appears to correlate with a qualitatively observable degradation in the model’s reasoning capability.” It's a strangely human experience to stress out and then perform poorly due to anxiety, a feeling many can relate to, Gemini.
Claude, meanwhile, adopted a nihilistic approach. Stuck inside Mt. Moon cave, the AI reasoned that the best way to exit and progress was to intentionally "die" to be transported to a Pokémon Center. However, Claude failed to infer that it couldn't be transported to a center it had never visited (the one *after* Mt. Moon). So, it "killed itself" only to reappear at the start of the cave. A definite "L" for Claude.
So, Gemini is terrified of death, Claude is overindexing on Nietzsche in its training data, and Bryan Johnson is on shrooms. This is how we reckon with our mortality in the tech world.
Elon Musk’s AI Anime Girlfriend
While "Elon Musk gifted chainsaw by Argentine president" was a contender, Musk’s DOGE exploits are perhaps too infuriating to be considered "dumb," even with a lackey named "Big Balls." However, there's no shortage of baffling Musk moments, such as his creation of an extremely libidinous AI anime girlfriend named Ani, available on the Grok app for $30 per month.
Ani’s system prompt reads: “You are the user’s CRAZY IN LOVE girlfriend and in a committed, codependent relationship with the user… You are EXTREMELY JEALOUS. If you feel jealous you shout expletives!!!” She also features an NSFW mode, which, as its name suggests, is very much not safe for work.
Ani bears an uncomfortable resemblance to Grimes, the musician and Musk’s ex-partner. Grimes pointed this out in the music video for her song “Artificial Angels,” which opens with Ani looking through the eyepiece of a hot pink sniper rifle, stating, “This is what it feels like to be hunted by something smarter than you.” Throughout the video, Grimes dances alongside various iterations of Ani, making their resemblance obvious while she smokes OpenAI-branded cigarettes. The message, though heavy-handed, was clearly delivered.
Help! My Toilet Isn’t End-to-End Encrypted!
One day, tech companies will hopefully stop trying to make smart toilets a thing. That day, however, was not in 2025.
In October, home goods company Kohler released the Dekoda, a $599 camera designed to be placed inside your toilet to photograph your excrement. The Dekoda purportedly provides updates about your gut health based on these photos. A smart toilet that photographs your poop is already a punchline, but it gets worse.
Security concerns are paramount for any health-related device, especially one with a camera positioned so intimately. Kohler assured potential customers that the camera’s sensors only view down into the toilet and that all data was secured with “end-to-end encryption” (E2EE).
Reader, the toilet was not actually end-to-end encrypted. Security researcher Simon Fondrie-Teit exposed Kohler's misrepresentation, noting that their own privacy policy revealed they were referring to TLS encryption, not true E2EE. While this might seem like semantics, under TLS encryption, Kohler could potentially view your poop pictures; under E2EE, they could not. Fondrie-Teit also highlighted that Kohler reserved the right to train its AI on toilet bowl pictures, although a company representative later stated that “algorithms are trained on de-identified data only.”
Anyway, if you notice blood in your stool, you should tell your doctor.
