As artificial intelligence advances at breakneck speed, tech workers are calling for a pause. However, can automated systems actually shut themselves down or is human intervention our only option?
When artificial intelligence leaps forward at breakneck speed, it raises a simple question. Can software turn itself off if things start getting dangerous? It’s the ideal question to bring up at the pub, at networking events and among freelancers.
The short answer, sadly, is no.
AI tools are essentially massive math engines. They follow instructions, spot patterns and process mountains of data. What they don’t have is a master switch for the world’s computer networks.
They simply can’t press an emergency stop button on tech development.
And that’s not even covering recent news of hackers hiding dangerous malware on a page hidden in Anthropic’s Claude.ai domain or Anthropic’s AI assistant Claude disobeying the company’s chief executive in a research scenario designed to test its willingness to follow instructions. The findings raise questions about human ability to rein in AI, The Bureau of Investigative Journalism reported. In a post sharing the research, Anthropic noted that although these weren’t real incidents, the study demonstrated “clear misaligned behaviour that should be studied further and mitigated”
Software cannot stop hardware
To understand why, we need to look at how these systems actually work.
An autonomous programme runs inside a fairly tight digital box. It might handle your inbox, draft code, or automate administrative tasks.
If it hits an error or runs into a safety boundary, it can drop the task it’s doing. It works a bit like a kettle shutting itself off when the water boils.
Stopping a single programme, though, is completely different from stopping global research. These systems don’t control the physical server farms or microchip manufacturing plants where future models take shape.
Even as software gets better at writing its own updates, it just pursues the goals humans gave it. It doesn’t have corporate authority and it can’t physically pull the plug on servers down the road.
“Each company—and country—is under intense competitive pressure not to unilaterally slow that acceleration.”
— From the ‘Pacing the Frontier’ statement
What the Pacing the Frontier open letter is really telling us
The conversation around safety went up a gear when the Pacing the Frontier statement came out. Over 1,200 tech workers at AI labs requesting global oversight tools signed an open letter.
Signatories include staff working at major labs like OpenAI, Anthropic, and Google DeepMind. They’re warning that progress could end up moving faster than human oversight can track.
Among the signatories are Jared Kaplan, Co-Founder and Chief Science Officer of Anthropic and Jakub Pachocki, Chief Scientist at OpenAI.
The heart of their concern isn’t sci-fi; it’s basic human competition. No single tech firm or nation wants to take their foot off the gas alone.
If one company pauses, a competitor takes the lead. That intense commercial drive creates a massive standoff.
Right now, is it a case where those who have built these LLMs lack the technical tools and commercial morality to coordinate a safe, industry-wide pause?
AI could help create a dramatically better future, but that outcome is not guaranteed. The world’s leading AI companies believe they could be close to automating AI research. It is hard to predict exactly how much this will accelerate AI progress, but there is a real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems.
To realize AI’s potential, industry, government, and society at large may need the option to buy time to address emerging risks, develop security measures, and strengthen oversight. But each company—and country—is under intense competitive pressure not to unilaterally slow that acceleration. And today, the world lacks the technical and governance tools to deliberately pace frontier-wide progress.
Building on work already underway to monitor frontier model releases:
We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”
“Signed because this statement helps establish common knowledge about the possible need for coordination mechanisms as automated AI research accelerates progress. I’d also like to see labs start designing these mechanisms voluntarily, even before the USG gets involved.”
John Schulman – Chief Scientist, Thinking Machines
I have worked in AI for a decade. Every few months in the last years, I have found myself surprised again and again at how rapidly the technology advances, no matter how often I update my predictions. AI may end up progressing far more quickly than societies and states can adapt. Thus, many prominent leaders in AI have publicly supported the ability to slow down, including DeepMind CEO Demis Hassabis.
At the same time, coordination is hampered by multiple factors, including competitive dynamics and antitrust regulations. I signed this statement (in a personal capacity) because I support greater *optionality* in the range of actions available, to ensure that states and societies around the world have time to thoughtfully prepare and adapt to this powerful technology.
Stephanie Chan, Staff Research Scientist, Google DeepMind
Why this matters to freelancers
You don’t need to work in Silicon Valley to feel the impact of this debate. Independent workers across almost every sector are already dealing with it on a daily basis.
Content creators and writers are already witnessing how automated tools are changing how they draft and edit, shaking up standard rates and what clients expect to pay.
Graphic designers, for example, are dealing with a double-edged sword. On the one hand, creative software can generate full visuals in seconds, but this in turn puts pressure on freelancers to deliver at breakneck speeds.
Software developers are seeing some of their tasks replaced with automated coding tools, which handle routine bug fixes. This is forcing developers to pitch for higher-level systems design.
Translators and admin pros are feeling pressure to bring down their fees. That’s because basic task automation handles routine work, yet their biggest selling point remains human nuance and deep expertise.
For most freelancers, the issue isn’t a sudden tech shutdown. It’s the sheer speed of change, where keeping up with new tools feels like a full-time job in itself.
In a previously reported article by The Freelance Informer, we explained how AI search for freelancers is changing fast. If your online presence still follows old-style search rules, you’re becoming invisible to the businesses that could hire you. Most self-employed people don’t know it’s happening. In this article, we suggest how to fix that AI-related problem.
Are Europe and the UK stepping in?
Governments are waking up to these challenges, though they’re trying very different strategies.
The European Union is taking a direct regulatory approach with its EU AI Act. The law requires safety checks, data audits and heavy fines for software labelled high-risk. Its aim is to force tech companies to prove their products are safe before launch.
The UK is taking a different route. Rather than passing one sweeping law, Britain relies on the UK AI Safety Institute.
The institute tests advanced systems alongside the developers who build them. The UK wants to spot risks early while letting the tech sector grow, but it doesn’t hold the legal power to order a hard pause on development.
How we could actually slow things down
If we want to pace technology, society does have a few real levers to pull.
For starters, governments could put hard caps on computing power. That would mean monitoring energy usage and restricting microchip sales to major research labs.
Second, workers themselves have power. Tech engineers and researchers can organise, refusing to build or deploy automated agents that lack proper testing.
Third, businesses and freelancers can vote with their wallets by demanding clear safety credentials before buying software.
Finally, public pressure can push lawmakers to broker real international treaties.
Will a slowdown actually happen?
What are the odds of seeing a deliberate pause anytime soon? Honestly, they’re pretty low right now. The commercial pressure behind tech companies is too strong.
Trillions of pounds in market value are driving massive investments. No firm wants to risk losing its edge by taking a break.
Add geopolitical rivalries into the mix, and voluntary pauses become even harder to pull off. Without binding international rules, commercial goals will keep driving development forward.
Control belongs to us
At the end of the day, artificial intelligence can’t set its own boundaries. Algorithms won’t act as global referees; they aren’t going to turn off competing data centres.
Pacing tech progress comes down to human decisions via international agreements, sensible laws, proper oversight and, at all costs, commercial boycotting or even sanctions. The technology will keep moving fast, but deciding how fast remains in our hands.
If you work in this field and have a story or solution to share, message The Freelance Informer LinkedIn Page or leave a comment.