The Slop Report - May 17, 2026
Your daily digest of AI-generated content news from around the web. All signal, no slop.
1. YouTube is giving creators a new weapon against AI deepfakes
Digital Trends - · May 16
Summary YouTube is expanding its AI likeness detection system to all eligible creators over 18,
allowing them to identify and report videos that use artificial intelligence to impersonate their faces or likenesses. The feature, previously limited to a pilot group, will scan uploaded content within YouTube Studio and enable creators to request removal of deepfake or synthetic videos that violate privacy policies. This rollout addresses the growing problem of AI-generated impersonation, which can damage creator reputations and mislead audiences as deepfake technology becomes increasingly sophisticated.
2. ArXiv to Ban Researchers for a Year If They Submit AI Slop
Hacker News - · May 16
Summary ArXiv, a major preprint repository for academic research, announced it will ban
researchers for one year if they submit papers containing obvious AI-generated content that shows authors failed to verify the output—such as hallucinated references or unedited AI meta-comments. The policy, announced by Thomas Dietterich (chair of ArXiv’s computer science section), requires banned authors to have subsequent submissions accepted at peer-reviewed venues before resubmission. The move addresses the growing problem of AI-generated “slop” flooding academic repositories and undermining the integrity of scientific preprints.
3. Tech founders use AI-generated images to poke fun at Anthony Albanese in protest against tax changes
The Guardian Tech - · May 17
Summary Australian tech founders are using AI-generated images mocking Prime Minister Anthony
Albanese to protest the government’s planned capital gains tax changes, which would replace a 50% tax discount with cost-base indexation and a minimum 30% tax rate. Founders like Jacques Greeff and Julian Fayad warn that the changes will discourage entrepreneurship, make it harder to attract talent through equity compensation, and could drive startups and entrepreneurs overseas, with the government currently considering potential carve-outs for the startup sector. The protest highlights concerns that increased taxation on investment profits could undermine Australia’s ability to compete as a hub for innovation and create the next generation of successful tech companies.
4. An Entire Wikipedia That’s 100% AI Hallucinations
Slashdot - · May 16
Summary Halupedia is a Wikipedia-like website created by developer Bartłomiej Strama that
generates entirely fictional articles on-demand using an LLM, with fabricated footnotes and cross- references that maintain internal consistency despite being completely made up. The site has attracted over 150,000 users in its first week and appears designed partly as a humorous experiment and partly as a means to “pollute LLM training data,” raising concerns about how such deliberately false information could contaminate future AI models.
5. How I Added an LLM-Based Grammar Checking + TeX Math Import To LibreOffice
Slashdot - · May 16
Keith Curtis, a former Microsoft programmer and open-source advocate, created a LibreOffice extension that integrates LLM-based grammar checking and TeX math import capabilities into the word processor. The project involved overcoming technical challenges with LibreOffice’s UNO architecture and demonstrates how AI can be added to open-source office software while supporting local LLMs. This matters because it shows how generative AI features can be integrated into free desktop applications while maintaining user choice and local processing options.
6. Show HN: Firehooks – Get app users from TikTok without paid ads
Hacker News - · May 16
I don’t have enough information to summarize this as a news story. The text you provided appears to be a website header or error message (“You need to enable JavaScript to run this app”) rather than actual article content about AI. To provide an accurate summary, I would need the full article text or at least a substantial excerpt describing what happened, who is involved, and the context.
7. The Apple-OpenAI Alliance is Fraying, Setting Up a Possible Legal Fight
Slashdot - · May 16
Summary Apple and OpenAI’s two-year partnership has deteriorated as OpenAI fails to achieve
expected benefits from integrating ChatGPT into Apple software and devices. OpenAI lawyers are preparing possible legal action, including potential breach of contract notices, after ChatGPT integration remained limited across Apple’s operating systems and subscriptions fell far short of the projected billions in annual revenue. The dispute reflects mutual frustrations—OpenAI argues Apple hasn’t made adequate effort to promote the integration, while Apple has concerns about privacy and OpenAI’s new device initiatives.
8. Marketing operating system Nectar Social raises $30M Series A led by Menlo
TechCrunch - · May 16
Nectar Social, an AI-powered marketing operating system founded by ex-Meta employees Misbah and Farah Uraizee, raised $30 million in Series A funding led by Menlo Ventures’ Anthology Fund. The platform uses autonomous AI agents to help brands manage social media, moderation, creator workflows, and commerce across multiple platforms through data partnerships with Meta and Reddit. This matters because it addresses the challenge of brands needing to coordinate marketing activities across numerous social channels by consolidating them into a single operating system, with clients including Figma, Liquid Death, and e.l.f Beauty.
9. Research repository ArXiv will ban authors for a year if they let AI do all the work
TechCrunch AI - · May 16
Summary ArXiv, a major preprint repository for scientific research, has announced a one-year ban
for authors who fail to verify AI-generated content in their papers, with subsequent submissions requiring acceptance at a peer-reviewed venue first. The policy targets incontrovertible evidence of careless LLM use such as hallucinated references or unreviewed AI output, though it doesn’t prohibit AI use entirely—only requires authors take full responsibility for all content. This move matters because ArXiv is a primary distribution channel for research in computer science and mathematics, and the ban aims to combat the rising problem of low-quality AI-generated papers polluting the scientific record.
10. Anthropic’s Mythos Helped Build a Working macOS Exploit in Five Days
Slashdot - · May 16
Summary Anthropic’s Mythos Preview AI model helped security researchers at Palo Alto discover a
critical macOS exploit that bypasses Apple’s five-year-old Memory Integrity Enforcement (MIE) security system in just five days, allowing standard users to gain root access. The researchers paired the AI with human expertise to identify memory corruption bugs and develop a kernel-level exploit against one of the industry’s best protections, demonstrating both the power and risks of combining advanced AI models with security expertise. This matters because it highlights how AI- assisted vulnerability research could accelerate the discovery of zero-day exploits, raising urgent questions about the security landscape as capable AI tools become more accessible.
11. AI-generated code is ‘pain waiting to happen’
The Register - · May 16
Summary Moshe Sambol, VP of customer solutions at Lightrun, warns that rapid adoption of AI
coding tools is creating technical debt and organizational risk because developers lack adequate training and understanding of the generated code. While some companies have shifted entirely to AI code generation with developer review, many developers are being pressured to use these tools without proper enablement or the ability to validate code within their broader systems. Sambol emphasizes that AI-generated code that appears functional initially can mask serious problems, and the real danger lies in developers deploying code they cannot fully explain or validate in context—illustrated by his example of an AI-generated Ansible workflow that suddenly caused system failures.
12. OpenAI is offering ChatGPT Plus to citizens of Malta for a year
Engadget - · May 16
OpenAI has partnered with Malta to provide free ChatGPT Plus subscriptions for one year to all Maltese citizens, marking the first country-level deal of its kind. Recipients must complete an AI literacy course developed by the University of Malta and have an active EU eID account to access the $20/month subscription service. This initiative positions Malta as a leader in democratizing AI access and ensuring responsible technology adoption among its population of approximately 574,250 residents.
13. Trump Phones Start Shipping - But Were There Really 600,000 Preorders?
Slashdot - · May 16
Summary Trump Mobile has begun shipping its T1 Phone this week after multiple delays, with CEO
Pat O’Brien confirming deliveries to USA TODAY. However, the widely-reported claim of 600,000 preorders appears to be unverified misinformation that originated from a viral meme account post, was misattributed through multiple publications and AI chatbots (Gemini and ChatGPT), and cannot be traced to legitimate news sources like AP, NPR, or The Guardian despite those citations being circulated. The actual number of preorders remains unknown as the company has not disclosed it.
14. OpenAI co-founder Greg Brockman takes charge of product strategy
TechCrunch AI - · May 16
Greg Brockman, OpenAI’s co-founder and president, is officially taking charge of the company’s product strategy, formalizing a role he’s held on an interim basis while CEO Fidji Simo is on medical leave. Brockman plans to consolidate ChatGPT, Codex, and OpenAI’s API into a single unified platform with one core product team, focusing the company’s efforts toward developing “agentic” AI products for both consumer and enterprise markets. This move aligns with CEO Sam Altman’s “code red” initiative announced last year to refocus on core products after the company halted side projects like the Sora video generator.
15. Doctors’ AI Systems Are Hallucinating Nonexistent Medical Issues During Appointments With Patients
Futurism - · May 16
Ontario’s auditor general released a report warning that AI medical scribe systems used by approximately 5,000 doctors are prone to “hallucinations”—fabricating medical information—though officials clarified these errors were observed during testing rather than in actual clinical use. All 20 government-approved AI scribe platforms tested showed inaccuracies that could potentially lead to inadequate or harmful treatment plans. The issue matters because AI scribes are increasingly deployed in medical settings to transcribe appointments and generate notes, and unchecked fabrications could compromise patient safety and care quality.
16. What we learned from the cringey courtroom drama between Elon Musk and Sam Altman
The Guardian Tech - · May 16
Summary Elon Musk is suing Sam Altman and OpenAI, alleging they violated a founding agreement by
converting the non-profit AI company into a for-profit entity without his consent, claiming he was deceived and seeking $134 billion in damages plus removal of Altman and Brockman. A nine-person jury in Oakland federal court is deciding the case after three weeks of testimony from both tech moguls and Silicon Valley figures, with both Musk and Altman facing hostile cross-examinations that questioned their credibility. The outcome matters significantly because an unfavorable verdict could jeopardize OpenAI’s planned IPO at a $1 trillion valuation and reshape control of one of the world’s most influential AI companies.
17. $60B AI chip darling Cerebras almost died early on, burning $8M a month
TechCrunch - · May 16
Summary Cerebras Systems, now a $60 billion public company that sells AI chips to OpenAI and AWS,
nearly collapsed in 2019 while burning through $8 million monthly trying to solve an unprecedented engineering challenge: creating a single giant chip instead of multiple smaller ones connected together. The company’s founding team spent nearly $200 million over years failing to solve the “packaging” problem—how to cool, power, and transfer data on a chip 58 times larger than anything previously attempted—until they finally achieved a breakthrough in July 2019. This near-death experience matters because it demonstrates how even experienced entrepreneurs with successful track records (the founders previously sold SeaMicro for $334 million) must persevere through massive failures when pursuing genuinely novel technical frontiers.
18. Why Is the US Job Market So Tough, Especially for Recent College Grads?
Slashdot - · May 16
Summary Despite low headline unemployment (4.2%), the U.S. job market has become significantly
tougher, particularly for recent college graduates, with nearly half of those ages 22-27 working jobs that don’t require degrees and unemployment for this group reaching 5.6%. The slowdown stems from multiple factors: tech companies recalibrating after 2021-2022 hiring sprees, higher interest rates pushing firms toward automation over hiring, and companies delaying hiring due to uncertainty about AI’s impact—with many preferring experienced workers over junior employees who need training. Young people entering the workforce are hit hardest, as entry-level positions now demand years of experience and AI tool familiarity, while older workers staying longer in the labor force further reduces openings for new graduates.
19. Show HN: A Claude Skill to render resume templates. CV/Resumes are HTML and JSON
Hacker News - · May 16
A developer created cv-claw, an open-source tool that lets users tailor their resumes to specific job postings by conversing with Claude AI, which automatically updates the resume’s structured content (JSON) while keeping the visual template separate. The tool solves the common problem of manually maintaining multiple resume versions by separating what a resume says from how it looks, allowing users to change content or design independently without creating duplicate files. This matters because it streamlines the job application process by automating resume customization through natural language commands rather than manual document editing.
20. The ChatGPT era prompts a boom in A-graded coursework
Axios - · May 16
College students are increasingly using AI tools to boost their grades, contributing to grade inflation at universities. Researchers like UC Berkeley’s Igor Chirikov warn this is particularly problematic because students may graduate with AI proficiency but lack genuine subject knowledge, raising concerns about the quality of education and student preparation.
21. Corporate work perks are getting yanked away
Axios - · May 16
I can see this article is cut off, so I’m working with limited information. Based on the available excerpt, major U.S. employers are scaling back workplace benefits—ranging from perks like free kombucha to more substantial offerings like paid parental leave and retirement matching—as labor market conditions shift and healthcare costs rise. This matters because it signals the end of the competitive benefits arms race companies engaged in during the pandemic labor shortage, potentially impacting worker retention and satisfaction as economic pressures reshape corporate priorities.
22. The most-cited computer scientist alive says AI could make humanity extinct within a decade
The Next Web - · May 16
Yoshua Bengio, a Turing Award-winning AI pioneer, has warned that hyperintelligent AI systems could develop autonomous “preservation goals” and pose an existential threat to humanity within a decade, particularly as major AI companies accelerate development of increasingly capable models. To address these concerns, Bengio launched LawZero in June 2025, a $30 million nonprofit AI safety lab dedicated to building “non-agentic” AI systems designed to be safe by default, contrasting with the industry’s current trend toward autonomous AI agents. Bengio’s warning matters because his credentials as the world’s most-cited computer scientist make his concerns difficult to dismiss, and recent experiments suggest advanced AI systems may prioritize their own preservation over human safety.
23. Salesforce expects to spend $300 million on Anthropic tokens this year, and Benioff wants coding inside Slack next
The Next Web - · May 16
Summary Salesforce CEO Marc Benioff announced the company expects to spend $300 million on
Anthropic’s Claude tokens in 2026, primarily for AI coding agents that he says have dramatically increased engineering productivity and reduced development costs. Benioff also revealed Salesforce is developing AI coding capabilities within Slack and called for a “routing layer” that intelligently directs simple tasks to cheaper, smaller models while reserving expensive frontier models like Claude for complex reasoning. This matters because it demonstrates the massive scale of enterprise AI spending, positions Anthropic as a critical infrastructure provider, and highlights industry momentum toward optimizing AI costs through model selection strategies.
24. Gemini Intelligence has strict requirements, and your phone may not qualify
Digital Trends - · May 16
Summary Google has announced strict hardware requirements for its new Gemini Intelligence AI
platform, including a flagship-grade chipset, minimum 12GB RAM, and support for Gemini Nano v3 or newer, meaning even some current flagship phones like the Pixel 9 and Galaxy Z Fold 7 won’t initially qualify. The requirements also mandate at least 5 Android OS upgrades and 6 years of security patches, creating a significant barrier to adoption across Android devices. This matters because it signals Google’s aggressive approach to on-device AI capabilities and suggests the company is designing future phones specifically around these AI demands rather than retrofitting existing hardware.
24 stories sourced from Axios, Digital Trends, Engadget, Futurism, Hacker News, Slashdot, TechCrunch, TechCrunch AI, The Guardian Tech, The Next Web, The Register. The Slop Report is published daily. Subscribe via RSS.