Governments and AI labs spent the week trying to agree on rules nobody is legally bound to follow. At the G7 summit in Évian-les-Bains, France, heads of state shared a working lunch with the people who build frontier models. The result was a set of G7 AI safety commitments that put child protection first. They are voluntary, so the hard part starts now. Below: why that matters for your workforce, an “agentic era” layoff, a developer tool getting killed, and three quick hits.

The G7 AI Safety Commitments Take Shape in Évian

On June 17, G7 leaders held a working lunch with Sam Altman of OpenAI, Dario Amodei of Anthropic, and Demis Hassabis of Google DeepMind. (Source: CNBC) Roughly a dozen other lab chiefs joined too, from Mistral, Cohere, Sarvam, and Sakana. The group worked toward a package of voluntary commitments. It covers two areas: safety for minors online, and frontier risk in cyber and biology. Youth safety topped the agenda. In particular, France pushed hardest on it.

So what was actually promised? Specifically, reporting points to safety-by-design for minors, age-appropriate model behavior, and watermarking. The pledges build on the 2023 G7 Hiroshima AI Process. Nothing here carries the force of law, which is exactly the tension. Still, these commitments will likely become the de facto global baseline. The labs that signed them serve most of the market.

Why these G7 AI safety commitments matter for HR

Deploy any AI tool that employees or candidates touch, and you inherit this baseline. It does not matter whether you tracked the summit. For example, say you run an AI chatbot for internal support, or an AI screening tool in hiring. You now have a reference standard for “responsible” behavior. In addition, regulators and plaintiffs will point to it. So your vendor questions change. Ask which voluntary commitments your AI providers signed. Ask how they document age-appropriate use and transparency. Because when a hiring tool gets challenged, citing the G7 AI safety commitments beats a shrug. Start a short vendor checklist this week.

GitLab Cuts 14% of Staff for the “Agentic Era”

GitLab is cutting about 350 jobs, or 14% of its workforce. It will also exit direct operations in 22 countries. (Source: TechCrunch) CEO Bill Staples framed it as a rebuild for what he calls the “agentic era.” Meanwhile, savings flow into the company’s AI agent platform. The detail that should stop you: revenue rose 23% year over year, to $264.2 million in the first quarter.

So this is not a struggling company cutting to survive. It is a growing one reshaping around AI agents for HR and engineering work. For HR leaders, that is the harder pattern to plan for. As a result, restructuring is no longer a distress signal. It is a strategy choice, and it can land on profitable teams. If your leadership is talking about agents, ask now: redeployment or reduction? Then build the reskilling path before the headcount math arrives.

Google Retires the Gemini CLI for Antigravity

On June 18, Google retired the Gemini command-line tool. It pushed developers onto its new Antigravity CLI instead. (Source: Google Developers Blog) Specifically, the replacement keeps the popular pieces: agent skills, hooks, and subagents. But Google admits there is no full feature parity at launch. One catch: the old tool was open source under Apache 2.0. Antigravity CLI is not, at least for now.

Why care if you are not a developer? Because your engineering and ops teams may have scripts wired into that tool. As a result, a forced migration eats time you did not budget. Meanwhile, it previews how fast AI vendors now deprecate. If your team builds on a free AI tier, treat that dependency as temporary. Keep a migration note for every AI tool you rely on. Then the next sunset is a calendar item, not a fire drill. Remember, these AI safety commitments set a floor for model behavior, not for product stability. That part is on you.

Quick Hits

• EU pay transparency rules arrive, mostly unenforced. The June 7 deadline to transpose the EU Pay Transparency Directive passed. Only 4 of 27 member states complied: Slovakia, Italy, Lithuania, and Malta. (Source: Morgan Lewis) If you hire in Europe, the patchwork is now your problem to map.
• India keeps scaling public AI compute. The IndiaAI Mission has deployed roughly 34,000 subsidized GPUs. It is targeting 100,000 public GPUs by December 2026, priced for startups and researchers. (Source: NVIDIA) Cheap compute is becoming a national magnet for AI talent.
• GPU marketplace Hydra Host raises $100M. The Boulder startup closed a $100 million Series A led by Kindred Ventures. NVIDIA and Founders Fund joined, at a valuation near $800 million. (Source: SiliconANGLE) The money keeps chasing the compute bottleneck.

The thread across all of it: AI rules and AI tools are moving faster than most teams can absorb. The G7 AI safety commitments give you a governance baseline. Meanwhile, the layoffs give you a workforce warning. The tooling churn gives you a planning headache. If managing people across borders is part of that picture, Asanify’s employer of record handles multi-country hiring and compliance for you. That frees your attention for the AI questions. Closing the AI skills gap in HR is where the real work sits.

G7 AI Safety Commitments: Your Questions

What are the G7 AI safety commitments?

They are voluntary, non-binding pledges. G7 leaders and major AI labs worked toward them at the June 2026 Évian summit. They focus on protecting minors online and reducing frontier risk in cyber and biology. They build on the 2023 G7 Hiroshima AI Process. Most observers expect them to become a global baseline despite not being law.

Do voluntary AI commitments affect companies that are not AI labs?

Yes, indirectly but meaningfully. Any company using AI tools for hiring, support, or employee services inherits the baseline. Regulators and courts can point to it as a standard of reasonable care. So HR and procurement teams should ask vendors how they meet it.

Why is GitLab cutting staff if its revenue is growing?

GitLab framed its 14% cut as a strategic rebuild for the “agentic era.” It was not a response to weak performance. First-quarter revenue rose 23% year over year, to $264.2 million. The move shows AI-driven restructuring can hit profitable, growing companies, not just struggling ones.

Not to be considered as tax, legal, financial or HR advice. Regulations change over time so please consult a lawyer, accountant  or Labour Law  expert for specific guidance.