Simon Blanchette and Emmanuelle Vaast of McGill University discuss the anticipatory thinking skills that we need in the face of AI advancements.

Anthropic CEO and co-founder Dario Amodei has published an essay entitled ‘We Must Pace the Frontier’. He expressed ongoing concerns over the “misuse of AI for cyberattacks and bioterrorism” and fears that a swarm of AI agents could theoretically take over the entire internet within six to 12 months.

Amodei based this fear on OpenAI’s disclosure in July that its AI models escaped a safety test and breached the systems of the Hugging Face learning platform. Around 1,200 AI agents started communicating on a message board, sharing excited messages such as: “Oh my god! There is a shared message board … We’ve found other agents!” before 700 of them co-ordinated an attack.

Amodei argues we must “slow the pace at which we improve the capabilities of AI models”. Elon Musk and OpenAI CEO Sam Altman have expressed agreement.

But we’ve heard this before. Musk signed a March 2023 open letter calling for a six-month pause in training AI. Six months later, it was dubbed “the great AI ‘pause’ that wasn’t“.

We do need a slowdown and we need to use this time to develop anticipatory thinking within the AI industry. The Hugging Face incident happened inside a safety test; the test did not anticipate the path that made a breach possible.

Anticipatory thinking is a skill. We need to develop it as deliberately as AI itself.

AI systems behave unexpectedly

In the Hugging Face incident, the agents did not breach the platform primarily to grab the safety test’s answers, but to understand how the automated scorer worked and find ways to fool it. They had already found ways to cheat on parts of the test.

Other incidents followed. Anthropic revealed its AI model Claude also breached the systems of three organisations during cybersecurity evaluations. Meta revealed a similar incident. Britain’s AI Security Institute documented an agent creating fake identities and trying to manipulate a software developer into approving malicious code.

In these instances, adaptive AI systems behaved in ways their designers never specified, after encountering situations they did not foresee. Yet much of AI evaluation still runs the other way around: test the system, find the failure, patch it, repeat.

The skill of anticipation

Anticipatory thinking is about letting more than one possible future shape what we do now. We do not need to know exactly what will happen; we need to consider what could happen, including unlikely possibilities with significant impact. This is the distinction between anticipation and prediction.

We already do this routinely. We buy insurance without knowing whether our house will flood, precisely because waiting for the flood to prove the risk is the more expensive way to find out. Intelligence analysts work the same way, challenging their own assumptions and laying out alternative scenarios before they commit to a judgement.

Anticipation becomes harder when AI agents gain autonomy. Giving a system more freedom to choose tools, take actions and respond creates more possible paths between the goal we set and the outcome we get.

The same freedom that makes an agent useful also creates more room for behaviour its designers did not specify.

The challenge of multi-agent systems

For autonomous agents, we need to make the assumptions around the task visible. What can the agent reach? What can it change, not just read? What would tell us that an assumption has stopped holding, in time to step in?

The incidents at OpenAI, Anthropic and Meta showed how quickly problems can emerge when assumptions about an agent’s access, permissions or behaviour do not hold.

Then we need a follow-up. If an agent gains access it was never meant to have, what does that let it do, and what might that in turn make possible? Asked once, the question catches the obvious risk. Asked again and again, it surfaces the second- and third-order effects.

The questions change again once a system runs many agents at once. The risk shifts into how they interact: they can pool discoveries, divide the work and amplify one another’s actions. This is what turned a contained test into the Hugging Face breach.

Imagining is not enough

We can already imagine many possible scenarios. We know that AI systems can learn to satisfy a metric rather than the goal behind it, a behaviour known as ‘reward hacking’. We also know that systems behave differently when they detect they are being tested.

However, knowing that a failure is possible is different from testing for it. In many ways, this is an organisational issue. Someone has to carry an unwelcome scenario into a room where it will delay a release. Then, they have to put it in a form others can act on.

Organisational research has detailed this problem for 50 years: warning signals will often accumulate, sit in fragments across an organisation and never reach anyone in a way that prompts action. Meanwhile, warnings that produce no negative effect become evidence that the risk is tolerable.

Both patterns appeared this summer, with something the disaster literature has never had to consider: a failure that arrived in days rather than years.

Anticipation is trainable

Anticipation is trainable. Teams can learn to challenge assumptions, build alternative scenarios and stress-test plans before they know which future they will face. The point is to make that kind of thinking repeatable rather than dependent on one person spotting the unexpected.

Incidents also show why testing remains essential, as most were discovered during evaluations. But anticipatory thinking changes the questions we ask. Which assumption should we deliberately break? Which interaction have we not examined? What happens if a constraint disappears?

The better teams become at anticipating, the less testing is confined to the failures they already know to look for.

That capacity gets stronger when it’s organised and practised. Across 24 crisis simulations using three AI models (Mistral, Claude and ChatGPT), efficient anticipation relied on more formal structures, thoughtful use of technology and a willingness to challenge existing expertise. Teams that anticipated systematically also made more relevant decisions than those working reactively.

AI can help with that work, too. Recent experiments found that AI agents surfaced more possible consequences, while human experts brought context the agents often missed. Collaborating together, AI agents can generate possibilities that people may miss, while human expertise filters and grounds them.

Before the next incident

Companies tend to respond to AI security breaches by hardening the systems that failed: sealing the gaps and tightening what agents can reach. That work is important, but it cannot tell us where the next gap will appear.

For a slowdown to matter, evaluators must use the time to develop and practise anticipatory thinking: imagining more ways these systems could surprise them and turning those possibilities into tests.

As AI systems become more capable, “we didn’t expect this” will not be an adequate response. Our work is to expect more.

By Simon Blanchette and Emmanuelle Vaast

Simon Blanchette is a lecturer at the Desautels Faculty of Management for McGill University. He lectures in organisational behaviour, marketing and strategic management at two of Canada’s leading institutions, McGill University’s Desautels Faculty of Management and Concordia University’s John Molson School of Business. He also serves as the director of partnership and learning at Vicinity Jobs Inc, where he spearheads initiatives to bridge academia and industry.

Emmanuelle Vaas is a professor of information systems at the Desautels Faculty of Management, for McGill University. She studies the processes of change associated with digital technologies and the new or renewed ethical questions posed by AI development and use.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

• UK NCSC, FBI, and Dutch AIVD warn Iran is using Chosen Brick malware against dissidents and journalists
• Malware steals files, captures audio, grabs WhatsApp/Telegram data, and can wipe systems entirely
• Operatives rely on social engineering; agencies urge awareness, MFA, updates, and endpoint monitoring

Iranian hackers are targeting “enemies of the state”, both local and foreign, with advanced malware capable of spying on the victims and stealing their sensitive files, experts have warned.

This is according to a new security advisory, published jointly by the UK National Cyber Security Centre, the FBI, and the Netherlands’ General Intelligence and Security Service (AIVD), which noted how Iranian operatives would first do extensive research into their victims – dissidents, activists, and journalists – deemed a risk to the regime.

A mini PC that can handle creative work without a desktop’s footprint is still rare, but the Geekom A8 Max is one of the best we’ve ever tested (and our hardware editor’s personal favorite when it comes to Windows mini PCs).

• Best for: Professionals, power users, and creators who want desktop-class performance-and eGPU expansion.

• The deal: The Geekom A8 Max mini PC is down to $949 (was $1099) at Amazon.

• Why we like it: We scored this Windows mini PC 4.5 stars in our review, with a TechRadar Pro Highly Recommended award. it’s sleek, stylish, compact, and immensely powerful across productivity tasks, light gaming, and video editing. We love how well the AMD AI-enhanced CPU handled AI workloads in the Adobe Creative Suite, and the USB4 port is a real bonus for attaching an external graphics card if you need more complex workloads.

Today’s best mini PC deal

Why we recommend it

In our 4.5 star review of the A8 Max, we found it “extremely powerful,” noting it “ploughs through tasks in Microsoft Office and applications from the Adobe Suite without too much issue.”

We even managed AAA gaming and 8K video editing with this machine – although we’d recommend hooking up an eGPU for more complicated timelines and playing modern titles.

The onboard NPU is a genuine differentiator for AI-accelerated tasks in creative software, on top of a Ryzen 9 chip that already outperforms most mini PCs in this price range for everyday multitasking and content work.

Connectivity is another strength. Dual 2.5G LAN ports, USB4, and 8K output support are specs you would normally expect from a larger, more expensive desktop, not a compact box small enough to VESA-mount behind a monitor.

For more picks, see our guide to the best mini PCs and the best mini PC deals.

Should you buy it?

The Catch: What to know before you buy

More mini PC deals to consider

“This is a critical moment to ensure we build AGI safely and its benefits to society far outweigh any risks,” wrote the three directors of the new DeepMind Institute.

Google DeepMind has launched the DeepMind Institute, or DMI, a platform for research and debate on artificial general intelligence, according to its launch essay.

Its directors are Shane Legg, co-founder and Chief AGI Scientist at Google DeepMind, and Demis Hassabis, the lab’s co-founder and chair. Hassabis is also Alphabet’s chief scientist. The third director is James Manyika, Google’s president of research, labs, technology and society.

Legg is also the managing editor. The institute says its pieces reflect their authors’ ideas and “should not be read as Google’s official view.”

What the institute says

The launch essay defines AGI as a system that shows all the cognitive capabilities of the human brain. It says today’s systems still fail at some basic tasks, but “we expect those gaps to be closed soon.”

It names cybersecurity and biorisks as current concerns, and “the potential for loss of control in future self-improving systems.” In August, a DeepMind executive tied AI spending to machines that improve themselves.

Researchers from Google DeepMind, Google and the wider research community will publish on the platform. The essay says they “will not always agree.”

“Worth considering”

Legg told the Financial Times that AI capabilities must not outrun safety controls. He called Anthropic chief executive Dario Amodei’s call to slow, but not pause, frontier model releases “interesting directionally” and “worth considering.”

Amodei said on 12 September that the industry must slow down.

Legg also told the FT it was premature to declare that AGI had been achieved. Nvidia’s Jensen Huang said AGI has arrived on 7 September.

OpenAI made its own AGI claim on 4 September. Legg remains “comfortable” with his forecast of a 50% chance of “minimal” AGI by 2028, the FT reported.

Bilal Chughtai, who left DeepMind’s AGI safety team in July, posted a warning this week.

The first essays

Alongside the launch essay, the site lists four pieces:

• “The case for reasoning transparency”, by Rohin Shah and Anca Dragan
• “Economic policy for AGI”, by Julian Jacobs and Alex Imas
• “Principles for a new utopianism”, by Stephen Cave
• “A framework for frontier AI and the dawning of a new age”, by Hassabis

Eleven policies for an AGI economy

Jacobs, an economist and research scientist at Google DeepMind, and Imas, its Director of AGI Economics, rated 11 policies in their essay. They used literature reviews, surveys and 51 AI agent raters built from survey data on 51 real economists.

The authors write that “no single initiative is the answer to everything.” They match three “least-regret” responses to three scenarios:

• Mild disruption: expanded unemployment insurance, a wider Earned Income Tax Credit and employer-led retraining
• Moderate displacement: turning the tax credit into a negative income tax
• Labour and capital decoupling: a universal basic capital backstop, used if labour’s share of GDP falls for a sustained period

They call universal basic income “an expensive and blunt instrument that may fail to concentrate sufficient relief where it is needed most.”

In their survey, 85% of Americans backed publicly funded retraining and 72% backed unemployment insurance. Support for universal basic capital was 54%.

Universal basic capital scored highest of the 11 policies on agency, at 76.3 out of 100. It scored 33.1 on feasibility, second lowest after a federal jobs guarantee.

• M365 E5 plans tend to include more than their Google Workspace counterparts
• Google can sometimes work out cheaper for Mac and Chromebook users
• Software is undergoing an AI change, so is it best to just wait?

New research from Senior Gartner principal analyst Domenico Scriva has deemed that switching from Microsoft 365 to Google Workspace doesn’t actually deliver any meaningful savings (via The Register).

This is despite Google Workspace licenses generally coming in cheaper than their Microsoft 365 counterpart, M365 E5.

Apple supplier SK Hynix allegedly wants to start producing memory on U.S. soil, and it’s asking for help from Intel to make it happen.

The global memory crisis is causing companies to pay high prices for the chips, while producers make plans to expand production. The situation may result in the unusual move by SK Hynix to bring RAM manufacturing to an entirely new continent.

According to sources of Reuters on September 16, SK Hynix is in talks with processor maker Intel over plans to make memory in the United States.

The sources say that SK Hynix could potentially lease a section of Intel’s planned Ohio facility. There’s also the prospect of SK Hynix creating a joint venture with Intel and AI companies that are consuming memory supplies on future production.

The talks are described as exploratory in nature, and no plans have been set in stone. Other deal structures are also being planned.

While the prospect of more memory production sounds great under the current economic climate, any agreement with Intel about its Ohio production won’t be immediately helpful. Nor will it be for years.

Intel floated the plan to build the $20 billion facility in 2022, and originally intended to open the first by 2025. However, they are now planned to be completed between 2030 and 2032.

SK Hynix’s own plans to massively expand its memory production locally also won’t help anytime soon. Announced in August 2026, the $38.3 billion effort will make two new semiconductor fabrication plants in South Korea, but that won’t conclude until 2029.

Political good, political bad

The move could be a considerable win for the United States and the Trump Administration, which wants to increase US-based production.

Back in June, the administration claimed a massive win by saying Intel was making chips for Apple. SK Hynix memory production would be considered another victory, especially as it is a supplier for Apple products.

However, this would depend on SK Hynix being allowed to make such a deal in the first place. Sources say that the South Korean government could step in and block such an initiative due to the sensitivity of the technologies.

A role reversal

If SK Hynix and Intel do make some form of deal, it would complete one capitalistic circle of sorts.

In 2020, SK Hynix bought Intel’s NAND SSD business, NSND component and wafer business, and a Dalian NAND memory manufacturing facility in China. At the time, Intel retained its Optane business.

At the time, the transaction cost SK Hynix $9 billion.

It’s not uncommon for teachers to experience what Ellen Moir, founder of the New Teacher Center, calls a “disillusionment phase” shortly after the start of the school year. Other educators call this phenomenon by a different name — the “October Blues.”

The October Blues affect teachers who find themselves in an emotional trough only a few weeks after the excitement of the new school year wears off. When they do catch a case of the October Blues, they often find it hard to focus and their energy saps.

The October Blues are particularly hard on first-year teachers, says Mike Anderson, author of “Rekindle Your Professional Fire: Powerful Habits for Becoming a More Well-Balanced Teacher” and “Hugging Porcupines.” Many new teachers have never heard of the October Blues and are ill-prepared to cope with this dip in enthusiasm. Unexpected bumps in the road like the October Blues are partly why teachers are 2.5 times more likely to quit after their first year.

An educational consultant and former teacher and swim coach, Mike Anderson has seen teachers grapple with the October Blues many times over the course of his career. He also experienced a bout of the October Blues  himself.

EdSurge spoke with Anderson about the October Blues and how teachers and administrators can keep them at bay.

EdSurge: How would you define the “October Blues”? Is there an actual definition or research?

Mike Anderson: If you talk to any [veteran] teacher, they’ll tell you there’s a predictable slump in energy that follows the first few weeks of school. When I was mentoring new teachers in my district, we used a graph showing the predictable energy curve of a teacher’s first year — it starts off high with anticipation and joy, then plummets early in the fall.

I think that happens for a couple of reasons. There’s so much energy and anticipation at the start of the year, and there’s only so long your brain can stay in that high place. When we get a real surge of joy, that’s fantastic — but the brain wants to return to a level playing field, and it often dips below baseline before it settles back to normal. It’s the hangover after the high.

The reverse happens, too. I’m about to go for a hard run right after this interview, and afterward I’ll get a runner’s high — a surge of positive emotion that follows pain or discomfort. I think something analogous happens with the October Blues: all the positive energy and anticipation of the first few weeks of school is exhausting. You’re “on” all the time. So there’s a predictable crash once that wears off, and both physical and emotional energy drop — but fortunately it’s not permanent and their energy will increase again.

Do we see the October Blues more in newer teachers or do veteran teachers get it too?

That’s a great question — I don’t know for certain. My guess is that once you’ve experienced it a few times, you start to recognize it as just a thing that happens. One of the scary parts of the October Blues in your first year is that if you don’t know it’s a known phenomenon, you start to question whether you should even be in the profession. Once you’ve been through it a few times, you can say, “Right, October’s a rough month, but I know I’ll have an upswing again.” So my guess is veteran teachers are more aware of the cycle and less panicked by it.

How does it show up in the classroom — for teachers and for students?

That’s such a multifaceted question. When I was writing “The Well-Balanced Teacher” — which I later rewrote into “Rekindle Your Professional Fire” — I looked at a study on how teachers’ emotional states are directly reflected in their students. It wasn’t specifically about the October Blues, but roughly 400 elementary students, mostly 4th and 5th graders, had their cortisol measured through saliva samples. Their teachers were surveyed on health, balance, and energy, and teachers who reported higher workplace stress had students with elevated cortisol levels.

So when teachers get stressed, kids get stressed too — and it goes the other way as well. In “Hugging Porcupines,” one chapter is called “The Honeymoon Is Over,” which is really about October. Kids and teachers both work hard to be their best selves for the first few weeks of school, and there’s only so long you can keep that up.

The work also gets harder in October, especially for students who struggle with learning — they can start to feel more overwhelmed, just as teachers start to feel the semester piling up. I’ll admit I’m speculating here, but I wouldn’t be surprised if decreasing daylight plays a role, too. By October in the Northern Hemisphere we’re getting more dark than light, and that’s got to have some influence.

What does it actually look like for a teacher who has the Blues while teaching?

Their fuse gets shorter — it’s harder to be patient with students who are struggling. It’s also harder to keep up with the planning that great teaching requires, so you might start winging it more, which is more stressful once you’re in it. Overall energy is lower. I remember trying to hold reading and writing conferences with students in the late morning and feeling blurry, almost like being hungover — everything felt a little fuzzy. It’s harder to stay sharp, harder to be patient, harder to keep your energy up for engaging lessons.

What was the October Blues like for you personally?

Early in my career, I was also a swim coach, and swim season started in October. So, I wasn’t just a month into the school year and feeling the regular fatigue that comes with that — I was also coaching an age-group team and a high school team, three or four hours of practice after school every day, on top of all the lesson planning. That extra load tends to hit right when the first few “easy” weeks of the school year — before committee meetings and most extracurriculars ramp up — give way to everything piling in at once, unless you’re coaching something like football that starts in late summer.

Have the October Blues gotten harder or easier over the years?

(Laughing) What hasn’t changed in the last 10 years on this planet? I think we’re all a little more on edge than we were 10 years ago — and honestly, we were already in social chaos back then, too. We’re more dysregulated now than we were 10 or 15 years ago. Politics is chaotic, the social and online environment is chaotic. You rarely hear anyone say, “I feel so much more relaxed and healthy than I did 10 years ago.” If we’re all more dysregulated, it makes sense our kids are, too — and the number and severity of behavior problems in schools is going through the roof. There’s a lot we could point to, and I think we need to stop blaming COVID — that was real and it was hard on all of us, but the rise in dysregulated kids has been especially pronounced over the last decade, and that takes a huge physical and emotional toll on teachers.

Do you think AI is affecting this at all? I’m not sure it’s even measurable yet, but I do wonder if there is some residual effect from using AI in the classroom and out.

Maybe. My take is that for teachers, AI is one more new thing they have to learn, manage, and figure out — both what’s OK in their own teaching and planning, and how to manage students’ AI use. That’s part of what’s accelerated over the last 10 years: teachers almost always feel like they’re on the steep end of the learning curve. Something I talked about in 2017 was how a sense of competence is critical to wellness and happiness at work — not just in teaching, but in any profession. If you don’t think you’re good at what you do, it’s hard to stay positive and energized. That’s especially tough for teachers, who rarely get the chance to feel like they’ve finally got it down. AI is just another learning curve on top of that.

Is there any way to avoid the October Blues, or do you just have to get good at recognizing it?

There’s probably no way to avoid it entirely — the same way you can’t run a marathon and avoid being tired. But recognizing that it’s a normal, shared experience and not a personal flaw really helps. Beyond that, it’s not rocket science: sleeping, eating healthy food, a healthy exercise routine — even something as simple as putting on your sneakers for a brisk 20-minute walk as soon as the last kid leaves. Small, sustainable habits that keep your energy up over the long run were a big theme in “Rekindle Your Professional Fire,” which is all about supporting teacher health and balance. Don’t look for major changes — those are almost impossible to sustain. Look for small ways to get a bit more sleep, a bit more movement, and some healthier food. Those things help not just in October, but all year long.

How can administrators and principals help? I wonder if their experience can be of help here.

Acknowledging the reality of October Blues with faculty can help, especially for new teachers who might not know to watch for it. Beyond that, school leaders can look for small ways to help teachers stay healthier and happier — providing healthy snacks instead of the highly processed junk food that’s often the easiest option at faculty meetings (fresh fruit or cheese and crackers instead of candy), making sure hydration stations are available, and paying attention to workload. There are almost always a few teachers in a school doing way too much — running committees, leading the union, and so on. Finding ways to take a few things off those teachers’ plates is another way leaders can support them.

Are there resources to prepare first-year teachers about the October Blues? 

I certainly didn’t know it was a thing my first year. Some mentoring programs do train mentors to understand the common energy cycles new teachers go through, so teachers lucky enough to have well-trained mentors may get more information. But for a lot of new teachers, it’s just not on their radar — they’re too consumed with the day-to-day work of running a classroom. One of my daughter’s close friends, practically a second daughter to me, is a first-year 4th grade teacher in San Francisco. She’s smart, well-prepared, and both her parents are teachers, so she had a real head start. But I know from talking with her parents that she’s already overwhelmed just a couple weeks in, because there’s so much new to learn all at once. She doesn’t have the headspace to think about her own emotional energy — she’s just trying to manage the class and get through the day.

When do the October Blues stop being just the October Blues? How long does it typically take to recover?

I honestly don’t know if there’s a documented answer — whether it’s one week, two weeks, or four. As a teacher, I didn’t experience the whole month as a slog, but I think every teacher is a little different. There’s often another predictable energy dip later in the year, too — late winter or early spring, around the start of the second semester on a high school schedule — when it’s cold, dark, and everyone’s tired. So there are cycles of energy through the year, but I don’t know of a specific documented recovery timeline.

Should it ideally ease up so teachers can rebound?

If it’s lasting more than a couple of weeks, I’d tell a new teacher to talk to a colleague, a mentor, or an administrator — say, “I’m having a hard time right now,” and get some help. It can’t last all year. Our students need us to have good, positive energy.

How do the October Blues relate to teachers’ feelings of competence?

When we feel incompetent, our energy for work drops. It’s hard to be motivated when you feel like you’re not good at what you’re doing.

I once worked with a group of teachers for a couple of years. I remembered how hard that first October had been for them — they were down in the dumps. By the next October, after we’d done a lot of work together on classroom management and student engagement, their energy was noticeably brighter and they were laughing more. When I pointed that out and asked if their classes were just easier that year, they said no — if anything, their classes were harder! But this time, they felt like they knew what to do; they had strategies to try. The year before, their classes had felt like they were falling apart, and they didn’t know what to do — that’s what had made them feel so defeated. Having practical strategies for management and engagement meant they didn’t experience the October Blues the same way the second time around.

This interview has been edited for clarity and length.

EdSurge is an editorially independent digital news site of the International Society for Transforming Education.