🟡 Fear vs. optimism
Semafor Technology · Semafor Technology · 2026-06-05
Semafor Technology's newsletter reports that companies are struggling to measure AI's ROI, the NSA is using Anthropic's unreleased Mythos model for offensive hacking operations, and labor market data shows limited evidence of AI-driven job displacement despite growing public anti-AI sentiment.
Appears in
Extraction
Topics: ai-roiai-securityai-labor-marketvoice-aiai-investment
Claims
- Companies are reducing AI spending primarily because results are too difficult to measure, not because tools lack utility.
- The NSA is reportedly using Anthropic's unreleased Mythos model for offensive hacking, despite the model being shared selectively to help organizations build defenses.
- Cloudflare reports most internet traffic is now AI-generated, one year ahead of earlier projections.
- ADP and Indeed labor market data show little evidence of widespread AI-driven job displacement, with demand for software developers ticking up over the past six months.
- Jensen Huang's public endorsements of companies at industry events are driving stock price increases of up to 10%, analogous to the Trump bump phenomenon.
Key quotes
The question is, are we going to need everything we're building? — Gary Cohn, IBM Vice Chairman
The US National Security Agency is reportedly using Anthropic's cutting-edge Mythos AI model for hacking.
When we just compare what AI can do [in] 2022 and today, this impact should be bigger today if the technology were making software skills redundant. — Hisayuki Idekoba, Recruit Holdings CEO