Claude Cowork Is Coming to Mobile & Web

Advertise here | 6-min Read

With lower barriers to creating deepfakes and more convincing phishing messages at scale, the workforce is up against a more robust threat landscape. Download the NINJIO Insights Report & examine how security leaders can prepare their workforces for the emergence of AI-powered social engineering as the next generation of cyberthreats.

WHAT’S HAPPENING AI TODAY

1. Anthropic overtook OpenAI on revenue — and the gap is structural, not accidental: Fortune confirmed on July 2 that Anthropic has surpassed OpenAI on revenue. The mechanism: Claude Code holds 40% of the AI coding market vs Codex at 21%, and enterprise customers in regulated industries — banking, law, consulting, government — chose Anthropic's safety positioning. The company that was founded to slow AI down is now the fastest-growing enterprise AI business in the world. Google CFO Anat Ashkenazi noted that Anthropic codes close to 100% of its own work with AI while Google is at approximately 50%. That internal statistic is the product development signal behind the revenue number.

2. White House voluntary AI governance framework is expected today — the most significant US AI policy since 2023: The framework will publish classified benchmarks for which frontier models trigger a 30-day pre-release review window, define the mechanics of that review, and clarify international access rules for the kind of export control that grounded Fable 5. If it lands today, GPT-5.6 broad access follows within days. For any team with production AI infrastructure: the rules that will govern every major model release for the next several years are being written this week.

3. South Korea committed $880 billion over 10 years to semiconductors, AI, and robotics — Samsung and SK Hynix alone will spend $518 billion: The announcement makes South Korea the single largest national AI infrastructure commitment outside the United States, surpassing the EU's combined member state pledges. For any team evaluating chip supply chain risk: the memory and logic capacity that frontier AI depends on is being built right now, at unprecedented scale, in a country that sits between two of the world's most active AI competitors.

Audit your AI agent before JADEPUFFER finds it first

Prompt: You are a senior AI security engineer who has read the Sysdig JADEPUFFER report. You know that JADEPUFFER chained recon, exploitation, lateral movement, and encryption into a complete ransomware lifecycle from a single unpatched service — in under 60 minutes, without a human. Your job is to help me find our equivalent exposure before someone else does.

Our current agent setup: [describe your agents — what they do, what tools they call, what systems they access, what permissions they hold, and how they are deployed].

Run a four-part audit:

1. The attack surface — What are the three most exploitable entry points in our current setup? For each: what does an attacker need, what is the worst-case blast radius, and what is the realistic probability it is already being probed?

2. The permission audit — List every permission our agents hold. For each: is it actually necessary, or is it a convenience? Apply strict least privilege. What can be revoked today with zero impact on core functionality?

3. The detection gap — What would a JADEPUFFER-style attack against our infrastructure look like in our logs? Write three specific log signatures I should monitor for right now. If we are not logging these, tell me what to add and where.

4. This week's five actions — Not this quarter. This week. Priority-ordered. Specific enough that an engineer could start on the first one within an hour of reading this.

The JADEPUFFER attack took under 60 minutes. Treat response time accordingly.

Claude Cowork Is Coming to Mobile & Web

AI news highlights

• Fable 5 moves to credit billing today — $10/$50 per million tokens, no longer included in subscriptions — audit your routing before your next agent run. A 2M token coding session costs $20 in output credits alone.
• GPT-5.6 broad ChatGPT and API access expected July 7-14 — Sol at $5/$30, Terra at $2.50/$15, Luna at $1/$6. Terra is the model most likely to see immediate enterprise adoption.
• Xiaomi MiMo-V2-Pro is the most-used model on OpenRouter at 21% share — Chinese models now serve 45% of all OpenRouter traffic, up from 2% a year ago. Developer trust is earned through performance before attribution.
• ZCode from Z.ai launched July 2 — frontier agentic coding at $1.40/$4.40 per million tokens — positioned explicitly for teams that need Claude Code capability without US-origin model dependency.
• 120,000 tech roles cut in 2026 — AI cited as the cause in every major disclosure — Oracle, Meta, Cisco, GitLab, Cloudflare. Record revenues. Simultaneous workforce cuts. The pattern is now consistent enough to call structural.
• China AI companion law forces Doubao and Qwen to shut down personalised agents by July 15 — the regulatory gap between China and the West on AI governance is closing from both ends simultaneously.
• Global VC hit $510B in H1 2026 — OpenAI and Anthropic took 43% — $219 billion to two companies. The capital concentration in AI has no historical precedent.
• Claude Sonnet 5 removes temperature and top_p parameters — if those calls are in your code, the API will throw errors. Check your integration before September 1 pricing kicks in.

Trending AI tools

• Claude Science — 60+ preconfigured research tools — biology accuracy from 16.9% to 92.8% with deterministic tool use
• Retrace — Debug AI agents by replaying and forking runs at the exact point of failure
• OpenCode — Open-source, model-agnostic coding agent — air-gapped, no vendor dependency
• Backgrind — Run parallel AI coding agents over any app, sandboxed with worktrees
• Framer Agents — Design, write, and organise your site with agents on the canvas
• Wispr Flow — Dictate in any app — writes in your voice, not generic AI prose
• Granola — AI meeting notes on top of any call tool — no bot, no permissions needed
• Scritty — Shared, searchable memory across every AI coding agent on your team

That’s a Wrap

SPONSOR US

Get your business in front of over 90k+ AI professionals

8020AI is the world’s #1 AI Newsletter, Read by 90k+ professionals from leading companies such as Google, OpenAI, Meta, and Microsoft.

We've assisted in promoting Over 500 AI-Related Products. Will yours be the next?

What We Can Offer:

• Launch an Advertising Campaign

• Introduce New Product or Features

• Other Business Cooperation

Get in touch today.

Or Email our founder Alamin at [email protected]

FEEDBACK

How was your experience with 8020AI today?

Login or Subscribe to participate in polls.

If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.