Daily AI brief · June 1, 2026
GitHub Copilot's Flat-Rate Era Ends Today
GitHub Copilot flips to token-based billing today, Google retires Gemini 2.0 and teases a flagship 3.5 Pro drop this month, Colorado's AI Act is 29 days out, and a new DataGrail report shows two-thirds of your SaaS vendors are hiding their AI subprocessors from you.
1. GitHub Copilot's Flat-Rate Era Is Over — Effective Right Now *(GitHub Blog / MLQ.ai)*
As of today, all GitHub Copilot plans transition to usage-based billing. Instead of counting premium requests, every plan now includes a monthly allotment of GitHub AI Credits, with paid plans able to purchase additional usage.
Early community estimates suggest a single agentic coding session can consume $30–$40 in credits — three to four times a Pro subscriber's entire $10 monthly allotment.
The good news:
if you set your additional-spend budget to $0, Copilot simply stops when credits run out — you wait for the monthly reset or upgrade. You cannot be charged a cent over your subscription.
What it means for SMBs / MSPs: If you manage a team on Copilot Business or Enterprise, log into your GitHub admin console *today*, set per-user spending alerts, and cap overage before one power user running overnight agentic sessions drains the shared pool.
2. Google Retires Gemini 2.0 — And Gemini 3.5 Pro Is Incoming This Month *(Firebase Docs / Google AI Changelog / ofox.ai)*
Gemini 2.0 Flash and Flash-Lite models shut down today, June 1, 2026. Any apps still pointing at those endpoints need to update to a newer model like gemini-3.1-flash-lite to avoid service disruption.
On the upside,
Gemini 3.5 Pro was supposed to headline Google I/O on May 19 but didn't ship. Sundar Pichai told the audience "give us until next month" — meaning June 2026, with no committed date.
What did ship is Gemini 3.5 Flash, now generally available, billed as Google's most intelligent model for sustained frontier performance on agentic and coding tasks.
What it means for SMBs / MSPs: If your team or a client has any custom integrations calling the Gemini 2.0 endpoints, they are broken as of today — check app logs now. Meanwhile, hold off on locking new agentic workflows to the Gemini 3.x stack until 3.5 Pro prices and capabilities are confirmed.
3. New Report: 63% of Your SaaS Vendors Are Hiding AI Subprocessors *(Help Net Security / DataGrail — published today)*
145 AI-related laws were enacted by state legislatures in 2025, and more than 1,000 additional bills were introduced or revised, according to DataGrail's Privacy and AI Trends Report 2026.
The buried headline:
of 2,400 popular business software providers that advertised AI capabilities, 63.6% did not disclose third-party AI subprocessors in their legal documentation, exposing businesses to shadow AI risks they may not be aware of.
DataGrail also found that 32.8% of AI systems participate in at least one high-risk activity, including sensitive data processing and automated decision-making.
What it means for SMBs / MSPs: Your clients are inheriting compliance exposure from their SaaS stack without knowing it. A quick vendor AI-subprocessor audit is a billable, defensible deliverable right now — especially for clients in healthcare, finance, or legal.
4. Colorado's AI Act Is 29 Days Away — and Most SMBs Aren't Ready *(ost.agency / Wilson Sonsini)*
On June 30, 2026, Colorado's Senate Bill 24-205 takes effect — the first comprehensive state-level AI law in the U.S. If your business develops or deploys a "high-risk AI system" influencing consequential decisions about Colorado consumers — covering employment, housing, credit, healthcare, education, or insurance — you are covered.
Violations carry penalties of up to $20,000 each, enforced by the Colorado Attorney General.
What it means for SMBs / MSPs: If any of your clients use AI in HR screening, loan decisions, or patient triage workflows and have Colorado customers or employees, the clock is running. A one-page AI use policy and a documented risk assessment are the minimum starting point.
5. OpenAI Sunsets GPT-4.5 and Continues Quiet GPT-5.x Cleanup *(OpenAI Help Center)*
OpenAI o3 will be retired from ChatGPT on August 26, 2026 following a 90-day sunset period, and GPT-4.5 will be retired from ChatGPT on June 27, 2026 following a 30-day sunset period.
Meanwhile,
GPT-5.4 brings together advances in reasoning, coding, and agentic workflows into a single frontier model, incorporating industry-leading coding capabilities while improving how the model works across tools, software environments, and professional tasks involving spreadsheets, presentations, and documents.
What it means for SMBs / MSPs: Any client demos or internal tools hardcoded to GPT-4.5 will break on June 27. Do a quick API key audit and update model strings before then.
GitHub Copilot Admin Budget Controls *(Business / Enterprise plans)*
The new billing model introduces budget controls at three levels: individual users can set a monthly overflow cap in billing settings — setting it to $0 stops usage entirely once credits are exhausted with no surprise charges — and organization admins can set per-user budgets across teams.
- Who it's for: MSP techs managing developer clients, or any internal AI champion with a Copilot Business seat
- How long a first test takes: 15 minutes to pull usage reports and configure caps
- One concrete first step: Go to `github.com/organizations/[org]/settings/billing`, download the April usage report that's been available since May 12, and set your org-level additional-spend cap to $0 until you've modeled actual credit consumption. Raise it deliberately — don't leave it open.
If your developers use GitHub Copilot, set a monthly budget cap in the admin console today. GitHub Copilot's billing changed today, from counting premium requests to a monthly allotment of AI Credits. Admins can set per-user budgets, and a $0 overflow cap stops usage once credits run out, with no surprise charges. Look at last month's heaviest users first.
*For MSPs:* A one-hour cost governance session that baselines usage, sets caps and projects spend stops a surprise invoice before it arrives.
Don't recommend Gemini 3.5 Pro for new client deployments yet. It hasn't shipped publicly as of today —
Pro 3.5 is not generally available as of May 28, 2026
— and pricing is unconfirmed. Gemini 3.5 Flash is available and already competitive, but committing client infrastructure to Pro specs that haven't been published sets you up for a redesign when the real pricing lands.
AI pricing is leaving the subscription era and entering the cloud-metering era — and SMBs are the last to know it happened.
Today's GitHub Copilot shift is a signal, not an isolated event.
The controversy around Copilot's new billing is less a freak-out than a market correction. AI coding assistants are leaving the promotional phase and entering the managed-infrastructure phase, where winners will be judged not only by how much code they can produce, but by how predictably, governably, and economically they can help humans ship software.
The industry shipped more than ten major launches in a 22-day window in May alone; introductory rates are now the industry default, and teams that don't lock in workflows before promo periods expire should budget for 2–3× rate increases within 90 days of every major launch.
The practical implication: start treating AI tools the way you treat AWS — with budgets, tagging, and monthly spend reviews. The clients who build that muscle now won't be the ones explaining a surprise $800 Copilot bill to their CFO in September.