日本語
Sources: [1] [2] [3]
2026-08-16 Evening edition
AI News Daily

AI News Daily 2026-08-16

Evening edition
Retrospective roundup: notable AI developments from recent weeks, compiled this evening because no qualifying breaking stories emerged in the last 24-48 hours.
EU AI Act enforcement begins as OpenAI and Anthropic reshape free-tier access and flagship-model pricing.

Today's Highlights

Regulation & Policy Published: 2026-08-02
EU begins full enforcement of the AI Act, with fines of up to €15M
  • The European Commission announced on August 2, 2026 that full enforcement of the EU AI Act has begun.
  • Enforcement now covers bans on prohibited AI practices, transparency duties for certain AI systems, and rules for general-purpose AI (GPAI) models.
  • Violations can draw fines of up to €15 million or 3% of a company's global annual revenue, whichever is higher.
Why it matters Major AI providers such as OpenAI and Anthropic must urgently strengthen EU compliance programs, and Japanese companies offering AI services in the EU need to re-check their own regulatory readiness.
€15M
Maximum fine per violation
Source (Tier 1, European Commission): https://ec.europa.eu/commission/presscorner/detail/en/ip_26_1714 Source (Tier 2, CNBC): https://www.cnbc.com/2026/08/03/eu-ai-act-enforcement-powers.html
Model Release Published: 2026-06-30
Anthropic unveils flagship model Claude Sonnet 5
  • Anthropic announced its new flagship model, Claude Sonnet 5, on June 30, 2026.
  • Introductory pricing of $2 per million input tokens and $10 per million output tokens applied through August 31, 2026.
  • In August, that introductory pricing was confirmed to remain in place as the standard rate going forward.
Why it matters Changes to a leading model's performance and pricing directly affect enterprise AI cost planning, so companies using Claude in production should revisit their cost estimates under the confirmed pricing.
$10 per M
Output token price now standard
Source (Tier 1, Anthropic): https://www.anthropic.com/news/claude-sonnet-5
Model Release Published: 2026-08-06
OpenAI removes ChatGPT free-tier text chat limits, sets GPT-5.6 Luna as default
  • With weekly ChatGPT users surpassing 1 billion, OpenAI removed the message cap on text chats for free-tier users.
  • The new model GPT-5.6 Luna became the default model for both the free tier and the Go plan.
  • Paid plans additionally gain access to a higher-tier model, GPT-5.6 Sol.
Why it matters Easing free-tier limits is meant to reduce user churn to rival services, and it broadens the options available to individuals and small businesses relying on free AI tools.
1B+
Weekly ChatGPT users
Source (Tier 1, OpenAI Help Center): https://help.openai.com/en/articles/20001354-gpt-56-in-chatgpt Source (Tier 2, TechCrunch): https://techcrunch.com/2026/08/06/openai-brings-unlimited-chatgpt-text-chats-to-free-users/

Trend Overview

Wrap-up

Three things to remember

  • The EU AI Act is now enforced, with fines up to €15M or 3% of global revenue.
  • Claude Sonnet 5's introductory pricing ($2 in, $10 out per million tokens) is now the standard rate.
  • ChatGPT's free tier now offers unlimited text chat, defaulting to GPT-5.6 Luna.

What to watch next

  • How AI providers adapt compliance programs as EU AI Act enforcement matures.
  • Whether rival labs match OpenAI's expanded free-tier access.
  • Further frontier-model pricing moves as enterprise cost pressure grows.
This is a retrospective roundup of recent AI developments; no new stories met this evening's sourcing threshold in the past 24-48 hours.