AI News: Sora Dies April 26 as OpenAI Kills Its $15M-a-Day Video App

OpenAI kills Sora April 26, Snap cuts 1,000 jobs as AI writes 65% of code, PwC says 20% of firms capture 74% of AI value, FBI flags AI cybercrime

Top Stories

Sora Shuts Down April 26 — Six Days From Now

OpenAI’s AI video generator Sora will go dark on April 26, with the API following on September 24. The app hit number one on the iOS App Store when it launched in September 2025, racking up a million downloads in its first week. Six months later, active users had cratered below 500,000, while the service was burning an estimated $15 million per day in compute costs.

The shutdown killed Disney’s planned $1 billion investment in OpenAI. Under the deal, Sora would have generated user-prompted videos featuring over 200 characters from Disney, Marvel, Pixar, and Star Wars. No money ever changed hands — the agreement was never finalized before OpenAI pulled the plug.

The strategic logic is straightforward: OpenAI is redirecting compute toward coding tools and enterprise customers as it prepares for a possible IPO later this year. Sam Altman made the call to sacrifice a flashy consumer product for margins. When your video app costs $15 million a day and fewer than 500,000 people use it, the math writes itself.

Users have six days to export their content. After April 26, videos and images will not be recoverable.

Sources: The Decoder, Variety, TechCrunch

Snap Cuts 1,000 Jobs — AI Now Writes 65% of Its Code

Snap laid off approximately 1,000 employees on April 15, roughly 16% of its 5,261-person workforce. CEO Evan Spiegel cited “rapid advancements” in AI as the primary driver, saying small teams using AI tools had already accelerated work across Snapchat+, ad platform performance, and infrastructure. The company also closed over 300 open positions.

The numbers paint a clear picture of where Snap sees its future: AI now generates more than 65% of new code at the company. The restructuring is expected to cut Snap’s annualized costs by over $500 million by the second half of 2026, with layoff-related charges estimated between $95 million and $130 million.

Spiegel described the moment as a “crucible” — Snap is squeezed between well-funded giants and nimble startups, and he is betting that a leaner, AI-augmented workforce can compete where headcount alone cannot. Departing U.S. employees will receive four months of severance, continued healthcare, equity vesting, and transition support.

Source: TechCrunch

PwC: 20% of Companies Capture 74% of AI’s Economic Value

A new PwC study of 1,217 senior executives across 25 sectors found that three-quarters of AI’s economic gains flow to just one-fifth of organizations. The top performers are generating 7.2 times more AI-driven revenue and efficiency gains than the average competitor.

The gap is not about who spends the most on AI. It is about how they use it. The leading companies are 2.6 times more likely to use AI to reinvent their business models rather than just cut costs. They are nearly twice as likely to deploy AI systems that operate autonomously within guardrails. And the single strongest predictor of AI-driven financial performance is not efficiency — it is industry convergence, using AI to expand beyond traditional sector boundaries.

Meanwhile, 56% of companies surveyed said they have seen no significant financial benefit from AI at all. The hype-to-value pipeline has a massive bottleneck, and most organizations are stuck on the wrong side of it.

Source: PwC

Quick Hits

  • FBI flags AI cybercrime for the first time: The FBI’s 2025 Internet Crime Report includes a dedicated AI section for the first time in its nearly 25-year history. Over 22,000 complaints referenced AI, with $893 million in losses. Scammers deployed voice clones, deepfake videos, and fake social profiles. Total cybercrime losses hit a record $20.87 billion — up 26% from 2024. The real AI figure is likely far higher, since most victims don’t realize AI was involved. FBI, The Register
  • Perplexity launches $200/month AI desktop: Perplexity’s “Personal Computer” feature hit Mac on April 16 for Max subscribers. It can search, read, and write local files, access iMessage, Mail, and Calendar, and run 24/7 on a Mac mini as an always-on agent. The cloud-based version for Pro subscribers ($20/month) launched in February. MacRumors
  • Tech layoffs pass 78,000 in 2026: Nearly 48% of the 78,557 tech jobs cut so far this year have been attributed to AI and automation, according to tracking data. The largest recent rounds came from Meta (8,000 planned for May 20), Snap (1,000), Disney, and Oracle. Tom’s Hardware
  • 600+ state AI bills introduced in 2026: State lawmakers have filed over 600 bills regulating AI this year. Indiana, Utah, and Washington have already enacted laws prohibiting health insurers from using AI as the sole basis for denying claims. The federal RAISE Act, imposing transparency requirements on frontier model developers, took effect March 19. Alston & Bird

Worth Watching

The AI layoff math is getting harder to ignore. Snap’s disclosure that AI writes 65% of its code — and that this justified cutting 16% of its workforce — is the most concrete data point yet linking AI capability to job elimination. Until now, companies have cited vague “efficiency gains” or “restructuring.” Snap put a number on it. Other companies will be asked to do the same, and the answers will shape how regulators, workers, and investors think about AI’s labor impact through the rest of 2026.

Sora’s failure is a signal about AI video’s economics. The product worked. Users liked it. It still lost $15 million a day. If OpenAI — with more compute than almost any company on Earth — could not make AI video generation economically viable, the path forward for competitors is narrow. The surviving AI video companies will either need to find dramatically more efficient architectures or charge enterprise prices that consumer users will not pay.