One-Line Summary

AI’s near-term impact on jobs looks more measured even as capital and markets double down, with OpenRouter’s funding and momentum stocks underscoring demand for practical deployment.

Big Tech

Altman says AI unlikely to trigger a jobs apocalypse

OpenAI CEO Sam Altman says AI is unlikely to cause mass job losses in the near term, revising his earlier concerns about white-collar displacement. Speaking at a Commonwealth Bank of Australia conference in Sydney on May 26, he said OpenAI was “roughly right” on the tech but “pretty wrong” on the social and economic impacts. ¹

Altman emphasized that a “human part” of many roles remains hard to replace, noting he tried letting AI reply to Slack and email before reverting to personal responses. He added he is “grateful” that his intuition about entry-level white-collar job losses being eliminated was off. ¹

While some companies have announced selective role reductions tied to AI, Altman said a jobs apocalypse is unlikely. Reuters also notes OpenAI is preparing to confidentially file for a U.S. IPO, situating his comments against intensifying investor interest. ¹

Industry & Biz

South Africa resets AI policy timeline after citation scandal

South Africa appointed an independent seven-member panel to revise a national AI policy after withdrawing a prior draft that contained fictitious and potentially AI-generated references. The government targets publishing a revised draft for public comment by January 2027 and aims to submit it to the Cabinet by November 2026; two officials are suspended pending investigation. ²

Officials said the withdrawal was necessary to restore credibility after internal checks failed to catch the flawed citations exposed by News24. For organizations, the episode is a reminder to disclose AI use in drafting and to verify all references in policy or external documents. ²

OpenRouter raises $113M and touts multi-model growth

TechCrunch reports OpenRouter, a unified gateway and API for large language models founded in 2023, raised $113 million in Series B led by CapitalG; The New York Times pegs its post-money valuation at about $1.3 billion. The service lets enterprises choose among 400-plus models from providers such as Anthropic, Google, OpenAI, xAI, and DeepSeek. ³

OpenRouter says it has 8 million users and processes about 100 trillion tokens per month (roughly 25 trillion per week), up fivefold from 5 trillion per week six months earlier. For teams, this supports a “multi-model” strategy to balance cost and quality without vendor lock-in. ³

AI rally powers record two-month momentum surge

Bloomberg reports the MSCI global momentum gauge has outperformed the MSCI All Country World Index by 17 percentage points since the end of March, on track for the strongest two-month outperformance on record in data back to 1991. The run continues despite worries about slower growth tied to the Iran war. ⁴

This concentration around perceived AI winners raises pressure on companies to show tangible AI gains or risk investor fatigue. For operators, that means linking AI projects to measurable productivity and margin improvements rather than hype. ⁴

Community Pulse

Hacker News (18↑) — Mixed: commenters question GDP gains and worry that winner-take-all incentives drive excessive, inflationary resource use, while some empathize with growth-at-all-costs behavior. ⁵

"This is the danger of a winner-take-all economy. They have no other but spend insane amount just to remains on top, even if it doesn't make economic sense short and medium term. There is no evidence AI will impact the GDP positively. To make things worse, so far, the use of ressources (electricity, chips) make it inflationary." — Hacker News ⁵

"To be fair tho, if you were in the shoes of "How to make the company grow as much as possible, make money as much as possible", ethics aside (because your goal would be just to make as much money as possible and score investments repetitively) , we'd probably behave in a similar way." — Hacker News ⁵

What This Means for You

The signal for knowledge workers is to design AI around, not over, human interaction. For communications-heavy workflows (Slack, email, support), keep AI in draft or suggestion mode and keep a human-in-the-loop to preserve trust and context—mirroring Altman’s own takeaway. ¹

OpenRouter’s raise and usage claims highlight real demand for choosing models by task, price, and reasoning depth. Even without an aggregator, you can emulate this by comparing outputs from at least two models in tools you already use, then standardizing which model to use for which task. ³

South Africa’s policy reset is a governance lesson: disclose when AI assists your drafting and verify every citation before publishing. Build this into your templates to reduce compliance and reputational risk with clients or regulators. ²

With momentum investors favoring perceived AI winners, budgets will increasingly flow to projects with clear, near-term productivity or margin impact. Tie each AI pilot to a baseline metric (time saved, conversion lift) and a go/stop threshold. ⁴

Action Items

1. Add an AI-use disclosure line to your templates: Insert a one-sentence note in external docs (proposals, reports) stating when AI assisted and that a human reviewed sources.
2. Compare two models on the same task: Run the identical prompt in two tools you already use (e.g., ChatGPT and Claude/Gemini) and record differences in clarity, structure, and speed.
3. Switch AI from auto-send to suggest-only for messages: For email/Slack that carries your name, keep AI drafting on but require human approval before sending for three days.
4. Create a simple ROI scorecard for one AI pilot: Define two metrics (e.g., minutes saved per ticket, error rate) and log this week’s baseline to decide whether to scale or stop.