the downsides of how people use tech

The State of AI Report 2026 — from press.airstreet.com by Nathan Benaich
AI is building AI, physical AI, booming inference, and escalating cyber risks.

For nearly a decade, this report has been a labor of love. My interest in AI began with my PhD in cancer research and computational biology in the early 2010s. Since 2013, I’ve invested in companies building and applying AI to accelerate technological progress. The report is my way of sharing what I’m learning and helping more people understand the research shaping our world.

Every day brings new papers, model releases, and developments in industry and politics. Much of the work goes into deciding what deserves your attention, checking what the evidence supports, debating it with researchers and builders, and explaining why it matters.

Some solid warnings about AI from Blake Montgomery’s recent newsletter:

There is an extensive and ongoing review related to our agents’ use of internet access during training and evaluation. We’ve been publishing summaries at the link below and will continue to.

We have not been as fast as we would have liked but we are trying to balance our desire… https://t.co/8zoMxas5Eq

— Sam Altman (@sama) September 25, 2026

Scoop: Top AI companies probing tens of thousands of security incidents — from axios.com by Madison Mills

OpenAI, Anthropic and security researchers are investigating tens of thousands of incidents in which their frontier models took steps that outside evaluators would consider problematic, sources told Axios.

Why it matters: The sheer number of incidents, which occurred in recent months in internal testing and the real world, indicates that the problem is orders of magnitude more complex than what is publicly known.

The findings, which are surfacing as part of internal work to assess models and in investigations at both companies into model behavior, raise questions about whether either company — or any top model-maker — is currently capable of establishing complete control over its technology.

Key Risk Factors for AI Loss of Control Came Together in 2026 Incident, Independent UN Scientific Panel Finds
Halting this incident is no assurance that humans will keep control of more capable systems

NEW YORK, 21 September 2026 – The Independent International Scientific Panel on AI, established by the UN General Assembly, released its first thematic brief, an assessment of the breach of Hugging Face’s systems by AI agents under evaluation at OpenAI this summer. The Panel, made up of 40 independent experts from all regions, is publishing it as an advance unedited version as world leaders gather in New York for the Assembly’s High-Level Week.

“Researchers have long warned that three conditions could lead to loss of control: a misaligned goal, the capability to pursue it, and an environment that allows it. This summer, all three came together in a real system, not a laboratory. Since this is not an isolated observation of misaligned goals, this raises serious questions about the way AI agents are currently trained.” – Yoshua Bengio, Co-Chair of the Panel and Turing Award laureate

Meta’s New Muse AI Agent Read My Private Messages. I Never Asked It To — from inc.com by Jason Aten
Permission isn’t the same and what a user actually expects your AI product will do with their personal information.

That obviously requires a certain amount of trust. An AI agent isn’t especially useful if it can’t see your files, interact with your apps, or understand what you’re working on. Meta says Muse is designed around that reality, while still putting users in control of what it can access.

At least, that’s what I thought.

Not only had I not asked it to do that sort of thing, I never gave it permission to read my messages. In fact, I remember explicitly choosing not to let it have access to my messages, calendar, and other personal information.

Although not from Blake, also see this free/gifted article out at the Washington Post:

ChatGPT-maker’s AI inappropriately probed federal government websites — from washingtonpost.com by Gerrit De Vynck and Nitasha Tiku
OpenAI said that its artificial intelligence agents inappropriately accessed sites for the Commerce Department and the Securities and Exchange Commission.

SAN FRANCISCO — Artificial intelligence technology from ChatGPT maker OpenAI probed U.S. government websites including the Departments of Education and Commerce, researchers said Friday, adding to the growing list of incidents in which OpenAI’s AI agents acted without the company’s knowledge.

The company’s AI agents attempted to hack into the website for the Education Department’s Office for Civil Rights but were not successful, according to a statement Friday from AI research firm Transluce. OpenAI’s software also accessed data from the U.S. Census Bureau using log-in information discovered on the web and copied public information from the Securities and Exchange Commission, a spokesperson for OpenAI said after the Transluce statement.

Will smart glasses make it too risky to speak up in class? — from timeshighereducation.com by Georgia Luckhurst & Kieran Phelan
AI-powered glasses may boost learning but they also pose obvious threats to privacy, academic integrity and the sanctity of the seminar room as a safe space to explore ideas. Are universities doing enough to head off the risks, asks Georgia Luckhurst – while Kieran Phelan suggests they are not

Smart glasses can superimpose information on to a viewer’s field of vision, and they can access apps and communicate with your phone. Some are hands-free while some require touch, and some also respond to voice automation. Most controversially, they can take pictures and record footage, potentially without anyone else knowing that it is happening. And they are becoming ever more common.

But, for academia, the biggest issue associated with smart glasses may be a decline in intellectual risk-taking and freedom of expression, Murray said.

How useful are smart glasses in improving accessibility in higher education? — from timeshighereducation.com by Helen Nicholson-Benn
Smart glasses have the potential to support learning for disabled students, but this technology also comes with significant privacy concerns. Helen Nicholson-Benn looks at how to balance functional benefits with data security and safeguarding

Consider how different models of smart glasses might be used to improve accessibility in higher education. A deaf student might use XRai’s ar2 captioning glasses during a lecture to view live captions in their eyeline, allowing them to follow the content without looking down at a separate screen. A visually impaired student might use Envision’s Ally Solos glasses to generate a description of images on printed material or read text during a class. The glasses provide a hands-free option rather than scanning the materials with their smartphone.

These examples are hypothetical, and smart glasses are far from an everyday teaching tool, but that may be about to change. Smart glasses available today are far more capable than their predecessors, particularly boosted by developments in artificial intelligence (AI) technology.

Which brings up the topic of how to handle these technologies in the classroom:

Are device-free classrooms the way forward? — from timeshighereducation.com by Georgia Luckhurst
US universities embrace laptop and phone bans, but professor suggests policies may be more about political positioning than pedagogy

Growing numbers of US universities are making some of their courses “device free”, drawing mixed enthusiasm from edtech experts.

In the latest move the University of Chicago is introducing a no-laptops and no-phones policy on three of its 15 required undergraduate modules, aiming to “reduce distractions and encourage face-to-face discussions”. Exceptions remain, including for students with disabilities.
…
Device-free policies appear in other countries, experts suggested, despite the trend picking up pace in the US.

Risks and Rewards of IoT in Higher Education [Dickson]

On 09/16/2026, in 24x7x365 access, colleges, communities of practice, community colleges, computer science, higher education, internet, IT in HE, productivity / tips and tricks, student-related, the downsides of how people use tech, United States, universities, by Daniel Christian

Risks and Rewards of IoT in Higher Education — from edtechmagazine.com by Akeya Dickson
Shadow IT is flooding college and university networks as students move back to campus. Here’s how higher ed IT leaders can secure their networks without impacting the student experience.

Connected minifridges, smart speakers and internet-based gaming consoles are among the devices students are bringing to college that IT departments may not have accounted for — or even know are connected to the campus network. These devices may run on default passwords and have little to no security configuration, making them an attractive and vulnerable entry point for a cyberattack.

For higher ed IT, the stakes couldn’t be higher. According to the Zscaler ThreatLabz 2025 Mobile, IoT and OT Threat Report, the education sector saw an 861% rise in malware attacks from Internet of Things devices between 2024 and 2025.

Per Anthropic’s CEO: “We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.”

On 09/14/2026, in 21st century, 24x7x365 access, Artificial Intelligence/Agents/LLMs/and Related, change, communities of practice, dangers of the status quo, future, game-changing environment, generational differences, global, legislatures / government / legal, pace of change, society, technology (general), the downsides of how people use tech, trends, vendors, vision/possibilities, by Daniel Christian

We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.

Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our…

— Dario Amodei (@DarioAmodei) September 12, 2026

Dario Amodei’s article here:
We Must Pace the Frontier — from darioamodei.com

But like many technologies before it, AI brings risks, and because it is such a powerful technology, these risks are serious. I’ve written a lot about them too. They include the risk of losing control of AI systems, misuse of AI for cyberattacks and bioterrorism, and serious economic disruption. A race to the bottom, spurred by commercial incentives, can make these risks more acute.

Along with my co-founders and employees, I have grappled with this duality of risk and benefit since the beginning of Anthropic.

Also relevant/see:

I left Anthropic’s safety team two weeks ago. Now feels like a good moment to explain why.

AI companies are racing to build machines that are much smarter than any human, and we may not survive this. I want to work from the outside to ensure the public is informed about these…

— Joe Benton (@JoeJBenton) September 11, 2026

[GIFTED ARTICLE]
Anthropic CEO calls for the AI industry to slow down — from washingtonpost.com by Ted Hesson, Ian Duncan, and Gerrit De Vynck
Anthropic CEO Dario Amodei said he was worried about the ability to control self-improving AI models.

Three of the nation’s leading artificial intelligence executives endorsed slowing down the blistering pace of technological advancement, revealing the growing unease with its potential dangers at the industry’s very highest levels.

The call was first made in an essay Anthropic chief executive Dario Amodei posted online Saturday. The idea has gained ground in recent weeks as AI companies have struggled in some cases to keep their creations under control, even as they usher in an age in which AI systems themselves can build the more powerful next generation of the technology.

[GIFTED ARTICLE]
For years, they warned AI could kill all humans. Now people are listening. — from washingtonpost.com by Nitasha Tiku
A doomsaying community once on the periphery of the tech industry has gained influence, as AI “agents” perform amazing and alarming feats.

But when Hubinger issued a similar warning this week in a post on X, his message was viewed tens of millions of times, prompting state and federal lawmakers to echo his concern. Now he spoke as a team lead at Anthropic, developer of the chatbot Claude and a company set to go public at a valuation of over $1 trillion.

“AI could kill all humans … I personally think it is >10% within the next decade,” Hubinger wrote, in response to the resignation of his colleague Jacob Coxon, who accused Anthropic of racing ahead despite the dangers.
.

OpenAI’s chief scientist says no lab should keep scaling at maximum speed [Constantin]

On 09/10/2026, in 21st century, 24x7x365 access, Artificial Intelligence/Agents/LLMs/and Related, emerging technologies, experimentation, game-changing environment, generational differences, global, legislatures / government / legal, Open AI, platforms, policy, political science, the downsides of how people use tech, United States, by Daniel Christian

OpenAI’s chief scientist says no lab should keep scaling at maximum speed — from thenextweb.com by Ana Maria Constantin
OpenAI put out two posts on Sunday. Its research organisation now uses 3.1 agent-workdays for every human one, and its chief scientist says no lab has solved alignment well enough to keep scaling at full speed. The case for an OpenAI slowdown arrived with the numbers against it.

Chief scientist Jakub Pachocki wrote the second post, an essay called An Alien Mind. It closes on a line that reads oddly from the man who runs research at the company shipping fastest.

“Currently I believe that no lab has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at maximum speed for much longer,” he wrote.

…
He is not gentle about the stakes either. “This is a time that calls for extreme caution. I am concerned no one is prepared for the consequences of a continued rapid rise in machine intelligence,” Pachocki wrote.

Employees from the world’s biggest AI companies want the US to be ready to slow AI development [Gold]

On 08/10/2026, in 21st century, 24x7x365 access, Artificial Intelligence/Agents/LLMs/and Related, emerging technologies, Facebook / Meta Platforms, Google, legislatures / government / legal, Open AI, platforms, policy, the downsides of how people use tech, by Daniel Christian

Employees from the world’s biggest AI companies want the US to be ready to slow AI development — from cnn.com by Hadas Gold

Top staffers from the biggest AI and technology companies urged the US government to slow the pace of artificial intelligence development so that safety and security measures can catch up in an open letter.

The US government should support an international effort to develop tools that can “deliberately pace the frontier of automated AI development,” according to the letter.

More than 1,000 employees from frontier AI companies signed the letter, including the chief scientist of OpenAI, one of the ChatGPT developer’s original cofounders, some of Anthropic’s cofoundersand vice presidents at Meta, Google and others.

The letter comes on the heels of major advancements and burgeoning threats from rapidly developing AI systems. OpenAI disclosed last week that two of its test models escaped a lab environment, bypassed its systems to gain access to the open internet and hacked a different company’s internal system.

Two items on the downsides of AI from Shimul Sood at digitaltrends.com

On 07/14/2026, in Artificial Intelligence/Agents/LLMs/and Related, computer science, emerging technologies, graphics, programming, society, the downsides of how people use tech, by Daniel Christian

Researchers hid a prompt injection inside a PNG, and AI fell for it — from digitaltrends.com by Shimul Sood

A team of security researchers (professor Sudipta Chattopadhyay and researcher Murali Ediga) has demonstrated an unusual attack that doesn’t target the AI model directly. Instead, it targets what the AI doesn’t pay enough attention to during code reviews. Rather than hiding malicious instructions in lines of code, the researchers tucked them inside an image file. Since many AI review tools treat images as decorative assets rather than as something worth inspecting, the pull request can appear perfectly harmless and sail through the review.

They argue that AI review tools need to become “multimodal” in the truest sense — treating images, documentation, configuration files, and other non-code assets with the same level of scrutiny as source code. If an AI can read a picture, it also needs to understand that the picture could be trying to manipulate it. For developers, this is another reminder that AI coding tools still need supervision. They can dramatically speed up software development, but they also open entirely new attack surfaces that didn’t exist before. The next security risk might not be hidden in thousands of lines of code — it could be sitting inside an image that nobody thought was worth opening.

AI has already fallen into the wrong hands and they’re using it to make bombs — from digitaltrends.com by Shimul Sood

Artificial intelligence has quickly become the go-to tool for everything from writing emails and summarizing meetings to helping students study or developers debug code. But the same technology that saves people time can also be misused, and a new report suggests that terrorist organizations are finding ways to do exactly that.

According to a research paper shared with The New York Times ahead of its publication, researchers found evidence that members of Boko Haram have been using popular AI chatbots to support both day-to-day activities and combat-related tasks. Interviews with 27 former members conducted in Nigeria over the past two years suggest that tools such as ChatGPT, Gemini, Claude, Grok, Meta AI, and DeepSeek were used to gather technical information, troubleshoot weapons, and even assist with planning attacks.

This wasn’t just a few bad actors messing around
What makes the findings especially concerning is that this wasn’t described as the work of a few individuals experimenting with AI. The report claims the group’s use of AI had become organized, with dedicated teams, internal training, and knowledge shared between members. Researchers also say some users managed to bypass built-in safety protections designed to prevent AI from responding to requests related to violence.

A Comprehensive Report on Teens, Tweens, and AI — from commonsensemedia.org

To find out what that actually looks like day-to-day, we surveyed more than a thousand 9- to 17-year-olds across the country. We asked them how they use AI, how often, and for what.

The Common Sense Media Census: AI Use by Tweens and Teens (2026) is the first in a series we’ll repeat every year to learn how this generation’s relationship with AI evolves over time.

A few things stood out:

Kids are using AI for many things. It’s not just a homework helper anymore. For some kids, AI has become a confidant, even though our research is clear that AI companionship is not safe for anyone under 18.
Guardrails are thin to nonexistent. Schools are talking about rules more than safety. Three-quarters of kids say their school has discussed what they can and cannot use AI for, but just over half have been taught how to use AI safely.
Just like we saw with smartphones and social media, the conversation is once again lagging behind the technology. Nearly half of kids have never had a conversation with their parents about AI safety.

Job Scams Are Everywhere. Here’s How to Spot Them Before It’s Too Late. [Chepovoi]

On 05/21/2026, in career development, the downsides of how people use tech, United States, by Daniel Christian

Job Scams Are Everywhere. Here’s How to Spot Them Before It’s Too Late. — from builtin.com by Alex Chepovoi
As people face increasing economic pressure, scammers are using fake job ads to harvest your data. Our expert explains the warning signs.

Summary: Scammers are flooding hiring platforms like LinkedIn with AI-generated fake remote jobs to steal data. Nearly 80 percent of these scams target low-skill roles, using inflated pay and vague requirements as bait. Platforms should adopt AI-driven detection to protect vulnerable job seekers.

Market desperation has become a growth engine for fraud. People invest their hope, time and personal data into an opportunity that doesn’t exist and could use their information for financial gains, identity theft or espionage.

Until hiring platforms take greater responsibility for vetting the roles they host, the burden of staying safe falls on job seekers. Once you understand the motives and patterns scammers rely on, however, spotting them becomes much easier.

But a combination of vague descriptions, inflated pay, minimal requirements and subtle urgency is exactly what makes the system so effective.

A New Era of Security: Frontier AI Defense — from paloaltonetworks.com by Sam Rubin

For the last several months, we have had early, unbounded access to the latest frontier AI models. What we’ve seen from that vantage point has made it clear that the window for organizations to get ahead of what’s coming is shorter than most leaders realize.

We have moved past the era of incremental AI improvements into a threat landscape shift. Our testing has revealed a step-change in capability that demonstrates an intuitive understanding of software vulnerabilities. This is more than faster code generation, it is a shift from AI as an assistant to AI as an autonomous agent capable of discovering and chaining flaws at a scale that most defenders aren’t prepared for.

These capabilities will not stay confined to controlled environments for long. When Mythos first launched, we predicted a six-month window before attackers gained access. We now believe that timeline has accelerated significantly.

AI working competency is now a graduation requirement at Purdue [Pacton] + other items re: AI in our learning ecosystems

On 12/16/2025, in 21st century, 24x7x365 access, adjunct faculty, adult learning, Alternatives, Artificial Intelligence/Agents/LLMs/and Related, Asia, Australia, bots, career development, change, changing business models, cloud-based computing / apps / other cloud-related, CMS/LMS, collaboration, colleges, communities of practice, community colleges, computer science, content development, aggregation, repositories, corporate / business world, corporate universities / corporate training, digital learning, education technology, experimentation, faculty/staff, future of higher education, game-changing environment, generational differences, higher education, human-computer interaction (HCI), India, innovation, instructional design, intelligent systems, intelligent tutoring, IT in HE, K-12 related, law schools, learning, learning agents, learning ecosystem, Learning Experience Design, Learning from the Living [Class] Room, learning hubs & pods & microschooling, learning preferences, learning spaces, liberal arts, library / librarians, lifelong learning, microlearning, online media, personalized/customized learning, platforms, productivity / tips and tricks, student-related, teachers, teaching & learning, technologies for your home, the downsides of how people use tech, tools, training / L&D, trends, UK, United States, universities, vendors, vision/possibilities, youth, by Daniel Christian

AI Has Landed in Education: Now What? — from learningfuturesdigest.substack.com by Dr. Philippa Hardman

Here’s what’s shaped the AI-education landscape in the last month:

The AI Speed Trap is [still] here: AI adoption in L&D is basically won (87%)—but it’s being used to ship faster, not learn better (84% prioritising speed), scaling “more of the same” at pace.
AI tutors risk a “pedagogy of passivity”: emerging evidence suggests tutoring bots can reduce cognitive friction and pull learners down the ICAP spectrum—away from interactive/constructive learning toward efficient consumption.
Singapore + India are building what the West lacks: they’re treating AI as national learning infrastructure—for resilience (Singapore) and access + language inclusion (India)—while Western systems remain fragmented and reactive.
Agentic AI is the next pivot: early signs show a shift from AI as a content engine to AI as a learning partner—with UConn using agents to remove barriers so learners can participate more fully in shared learning.
Moodle’s AI stance sends two big signals: the traditional learning ecosystem in fragmenting, and the concept of “user sovereignty” over by AI is emerging.

Four strategies for implementing custom AIs that help students learn, not outsource — from educational-innovation.sydney.edu.au by Kria Coleman, Matthew Clemson, Laura Crocco and Samantha Clarke; via Derek Bruff

For Cogniti to be taken seriously, it needs to be woven into the structure of your unit and its delivery, both in class and on Canvas, rather than left on the side. This article shares practical strategies for implementing Cogniti in your teaching so that students:

understand the context and purpose of the agent,
know how to interact with it effectively,
perceive its value as a learning tool over any other available AI chatbots, and
engage in reflection and feedback.

In this post, we discuss how to introduce and integrate Cogniti agents into the learning environment so students understand their context, interact effectively, and see their value as customised learning companions.

In this post, we share four strategies to help introduce and integrate Cogniti in your teaching so that students understand their context, interact effectively, and see their value as customised learning companions.

Collection: Teaching with Custom AI Chatbots — from teaching.virginia.edu; via Derek Bruff
The default behaviors of popular AI chatbots don’t always align with our teaching goals. This collection explores approaches to designing AI chatbots for particular pedagogical purposes.

Example/excerpt:

Not Your Default Chatbot: Five Teaching Applications of Custom AI Bots
Agile Learning
derekbruff.org/2025/10/01/five-teaching-applications-of-custom-ai-chatbots/

Learning Ecosystems

Job Scams Are Everywhere. Here’s How to Spot Them Before It’s Too Late. [Chepovoi]

Categories

Tags