Bête Noire

The premise

Maria is a recipe developer at a confectionery company whose orderly life cracks when Verity, a quietly menacing former schoolmate, joins as a new hire. Small things go wrong: a takeaway restaurant Maria swears existed never did, a colleague's name was always spelled differently, a past humiliation everyone remembers differently. Maria looks insane defending facts no one else disputes, because Verity is rewriting them. She wields a quantum computer that lets her edit reality and overwrite everyone's memories on a whim, gaslighting Maria into doubting her own mind. The episode is less about hardware than about who controls the shared record of what is true.

What the episode imagines

• 🔹❌Digital gaslighting
• 📱❌Social perception manipulation
• 👁️❌Corporate surveillance

How close are we in 2026?

No one has a quantum reality-editor, but the social mechanism the episode dramatizes, contested shared facts that no single person can verify, is arriving fast. The clearest proxy is generative AI quietly rewriting the record. NewsGuard's one-year audit, published September 2025, found the ten leading chatbots (ChatGPT, Gemini, Claude, Copilot, Grok, Perplexity, Meta AI and others) repeated false claims on news topics 35% of the time, nearly double the 18% rate of August 2024, after vendors wired in live web search and stopped declining to answer. Perplexity (47%) and Inflection's Pi (57%) fared worst. When the tools people now treat as arbiters of truth confidently assert falsehoods, Verity's trick of making everyone agree on a wrong fact stops feeling supernatural.

The gaslighting-as-fraud angle is already operational. In early 2024, engineering giant Arup lost HK$200M (about US$25.6M) when a Hong Kong finance clerk was talked into 15 wire transfers by a video call full of deepfaked colleagues, including a fake CFO. Deepfake-enabled voice-phishing attacks jumped 1,633% in Q1 2025; the FBI's 2025 report logged 22,000+ AI fraud complaints topping US$893M.

And the visual record itself is now editable by anyone. Google's Pixel 9 "Reimagine" tool, then rolled into Google Photos for everyone in 2025, lets a typed prompt insert convincing things that were never there, prompting Google to add invisible SynthID watermarks in February 2025. The shared past is becoming as malleable as the episode imagines, just without the quantum computer.

Key real-world developments

• Chatbots now get the news wrong 35% of the time

  NewsGuard's September 2025 audit found the top ten AI chatbots repeated false claims on news topics 35% of the time, up from 18% a year earlier, after adding live web search; Perplexity hit 47% and Inflection's Pi 57%.

• Deepfake CFO scams a firm out of $25M

  In 2024 engineering group Arup lost about US$25.6M when a Hong Kong employee joined a video call of deepfaked executives, including a fake CFO, and made 15 transfers, real-life gaslighting via synthetic colleagues.

• AI fraud losses now measured in billions

  Deepfake-enabled voice-phishing surged 1,633% in Q1 2025; the FBI's 2025 Internet Crime Report logged over 22,000 AI-related fraud complaints exceeding US$893M, with Deloitte projecting US$40B in annual US AI-fraud losses by 2027.

• Anyone can rewrite a photo's past

  Google's "Reimagine" tool, on Pixel 9 in 2024 and all Google Photos users by 2025, inserts realistic objects from a text prompt; Google began adding SynthID watermarks to edited images in February 2025 to flag fakes.

The verdict

The literal premise, a quantum computer that overwrites everyone's memories, is pure fantasy and isn't on any roadmap. But strip that away and the episode is eerily close in 2026. We already outsource "what's true" to AI systems that confidently fabricate, photo tools that let anyone edit the visual record, and deepfakes convincing enough to drain millions by impersonating people we trust. The missing piece isn't technology, it's a shared, verifiable source of truth, and that's eroding, not strengthening. Bête Noire's real subject, the weaponization of doubt, has already escaped fiction.

Sources

• NewsGuard: One-Year AI Audit Finds Models Spread Falsehoods 35% of the Time (2025)
• CNN Business: Arup revealed as victim of $25M deepfake scam (2024)
• Axios: Popular chatbots amplify misinformation (2025)
• TechCrunch: Google adds digital watermarks to Magic Editor / Reimagine images (2025)
• Deepstrike: Deepfake Statistics 2025 (vishing surge, FBI losses)
• Information Age (ACS): Google Pixel's realistic AI photos 'abandon reality' (2024)

Reviewed and updated by the How Close to Black Mirror editorial team on 20 June 2026. Progress scores are reasoned editorial estimates based on the cited sources, not scientific measurements.