AI News Archive: July 7, 2026 — Part 14
Sourced from 500+ daily AI sources, scored by relevance.
- Nigeria opens probe into Meta, X, and AI firms
On Techpoint Digest, we discuss Nigeria opening a probe into Meta, X, and AI firms, Zeepay's response to $11.6 million refund reports, Kenya's creation of new licences for Uber, Bolt, and Glovo, and Angola's decision to take Unitel public following a state takeover.
- Alef Education, Microsoft empower 25,000 educators across UAE through AI literacy training programme
The initiative marks one of the country’s most extensive educator-focused AI capacity-building efforts, reinforcing the UAE’s leadership in advancing AI-enabled education
- Dubai Chamber showcases digital opportunities at GITEX AI Europe
The platform serves as a single gateway for business services, offering integrated support to investors
- Total CX and Ibn Al-Nafees Hospital sign strategic partnership to advance AI-powered customer experience services
reflecting both organizations' shared commitment to enhancing customer experience through the integration of skilled talent and AI-enabled solutions
- Why AI could reshape South Africa's fruit export industry ?
AI represents the next evolution of that journey. The businesses that learn how to combine human expertise with intelligent technology
- Cisco outlines five essential strategies to close the AI trust gap and secure the agentic enterprise
As AI enabled threats move at machine speed, organizations must embed trust and security directly into their infrastructure to unlock the full potential of Agentic AI
- Electronic Caregiver Expands Access to Intelligent Care Across New Mexico While Advancing the Rio Grande Health Technology and Services Corridor
Electronic Caregiver Expands Access to Intelligent Care Across New Mexico While Advancing the Rio Grande Health Technology and Services Corridor USA Today
- What happens when your agent can touch money
Subscribe • Previous Issues Your CLI Was Built for Humans, Not Agents There’s a friendly debate among developers about how to give AI agents reliable ways to use external tools, data, and services so they can do useful work beyond generating text. One side favors CLIs, or Command-Line Interfaces, where agents run text commands against tools like Continue reading "What happens when your agent can touch money" The post What happens when your agent can touch money appeared first on Gradient Flow .
- The Sequence Knowledge #890: A Brief History of Model Distillation
The papers and techniques that laid out the ground work to evolve distillation to this level.
- Tools vs. Subagents: Building Effective AI Agents Without Over-Engineering
Tools execute code.
- Indian enterprises race to adopt AI but workforce readiness lags
Indian enterprises race to adopt AI but workforce readiness lags Techcircle
- AI accountability is now healthcare's next big challenge
AI accountability is now healthcare's next big challenge Healthcare IT News
- At MGB and elsewhere, AI enables new therapies sooner for patients
At MGB and elsewhere, AI enables new therapies sooner for patients Healthcare IT News
- UCSD aims to put AI research into practice
UCSD aims to put AI research into practice Healthcare IT News
- First robotic surgery network in Vietnam launched
First robotic surgery network in Vietnam launched Healthcare IT News
- How patients' mental health might benefit from robots
How patients' mental health might benefit from robots Healthcare IT News
- ECB tells banks: Fix AI cyber gaps by Oct. 31
The order covers the European units of JPMorganChase, Goldman, Citi and Morgan Stanley, and previews what U.S. regulators may eventually demand.
- Bowman: Low-risk AI usage should get lighter regulatory touch
Federal Reserve Vice Chair for Supervision Michelle Bowman said in a speech Tuesday morning that she is working with other regulators around the world to emphasize innovation in the banking sector, including with artificial intelligence.
- Register: Risky Future AI Tools for Underwriting ‘Demo Day’ on July 8
Insurance Journal’s Risky Future series is hosting the “AI Tools for Underwriting” Demo Day, a series of free AI tool demonstrations designed exclusively for Carriers, MGAs, Specialty Program Leaders, and of course, Underwriters and Claims professionals. This event spotlights AI …
- Hyundai showcases humanoid robot at FIFA World Cup in robotics push
Hyundai showcases humanoid robot at FIFA World Cup in robotics push Automotive News
- Cobots become simpler, smarter with AI
Collaborative robots have taken off in recent years as artificial intelligence makes them useful for a wider variety of applications.
- Kickbacks Takes An Outsider’s View While Bringing Ads To AI Agents
Andrew McCalip is a founding engineer at Varda Space Industries, where he oversees the manufacturing of things like hypersonic reentry vehicles and satellite buses (which are the central satellite bodies that carry a payload or instruments, not something that, like, shuttles satellites out to a launch pad). But he’s always had a taste for online […] The post Kickbacks Takes An Outsider’s View While Bringing Ads To AI Agents appeared first on AdExchanger .
- AI Might Not Need Perfect Data – Except When It Comes To Revenue
Every publisher and media company is asking the same question right now: How do we make AI work for our business? But in the rush to adopt, there’s a foundational question most teams aren’t asking loudly enough: Is your data actually ready for it? AI doesn’t conjure insights from nothing. It works on what you […] The post AI Might Not Need Perfect Data – Except When It Comes To Revenue appeared first on AdExchanger .
- 3 AI agents to improve marketing workflows
Pilot AI agents with a shared source of truth, practical KPIs, and human oversight that keeps strategy and quality on track. The post 3 AI agents to improve marketing workflows appeared first on MarTech .
- How AI discovery is changing everything marketers measure
AI-powered discovery is reducing the value of traffic-based metrics. Here’s what to measure instead to understand marketing performance. The post How AI discovery is changing everything marketers measure appeared first on MarTech .
- CEOs fear they’re underinvesting in AI
More than half of chief executives are concerned their businesses will fall behind due to limitations in technology foundations, according to a new survey.
- AI scammers are cloning voices and creating fake websites. Here is how to stay safe
AI scammers are cloning voices and creating fake websites. Here is how to stay safe Dallas News
- Attitudes toward data centers in Texas are shifting, and for good reason
Attitudes toward data centers in Texas are shifting, and for good reason Dallas News
- NexArt Adds Independent Timestamping and Confidential Mode to AI Execution Records
NexArt Adds Independent Timestamping and Confidential Mode to AI Execution Records azcentral.com and The Arizona Republic
- Influential Women Features Isabella Johnston: Redefining Talent Strategy With AI-Driven Workforce Alignment
Influential Women Features Isabella Johnston: Redefining Talent Strategy With AI-Driven Workforce Alignment azcentral.com and The Arizona Republic
- The Greatness Factory of Nashville and PCG Technologies Announce Strategic Partnership to Launch AI Clone
The Greatness Factory of Nashville and PCG Technologies Announce Strategic Partnership to Launch AI Clone azcentral.com and The Arizona Republic
- Couplr Becomes First Behavioral-Compatibility Advisor-Matching Platform Live Across 4 Major AI Ecosystems in Single Day
Couplr Becomes First Behavioral-Compatibility Advisor-Matching Platform Live Across 4 Major AI Ecosystems in Single Day azcentral.com and The Arizona Republic
- MCK Network Solutions and CORTAI Partner to Deliver Predictive AI Security
MCK Network Solutions and CORTAI Partner to Deliver Predictive AI Security azcentral.com and The Arizona Republic
- Insygna Publishes White Paper on Managing AI Agents as an Extended Workforce
Insygna Publishes White Paper on Managing AI Agents as an Extended Workforce azcentral.com and The Arizona Republic
- SPOTIO Launches Next Best Action: The First Predictive, AI-Powered Sales Optimization Engine Built for Field Sales
SPOTIO Launches Next Best Action: The First Predictive, AI-Powered Sales Optimization Engine Built for Field Sales azcentral.com and The Arizona Republic
- Zambuki Introduces Advanced AI-Powered Solution for Generating Leads for Cleaning Services
Zambuki Introduces Advanced AI-Powered Solution for Generating Leads for Cleaning Services azcentral.com and The Arizona Republic
- Global AI Industry Falls Short On Safety, Think Tank Warns
Global AI Industry Falls Short On Safety, Think Tank Warns Barron's
- AI Bill Keeps Growing as Alphabet, Amazon, and Meta Spending Is Set to Go Through the Roof
AI Bill Keeps Growing as Alphabet, Amazon, and Meta Spending Is Set to Go Through the Roof Barron's
- Palantir Looks Beyond the Pentagon With a New Commercial Win in Mexico
Palantir Looks Beyond the Pentagon With a New Commercial Win in Mexico Barron's
- Palantir Strikes a Deal With Mexico’s Top Insurer. Will It Boost the Stock?
Palantir Strikes a Deal With Mexico’s Top Insurer. Will It Boost the Stock? Barron's
- This AI Founder Admits He Was a 'Terrible Leader' — But a Simple Mindset Shift Solved His Biggest Problem
This AI Founder Admits He Was a 'Terrible Leader' — But a Simple Mindset Shift Solved His Biggest Problem Entrepreneur
- AI has not killed SaaS. It has killed bad SaaS
AI has not killed SaaS. It has killed bad SaaS Entrepreneur
- Malaysia’s Zetrix AI to Support the Philippines’ Public Blockchain under DICT Pact
Malaysia’s Zetrix AI to Support the Philippines’ Public Blockchain under DICT Pact Entrepreneur
- Your Brand Will Be Invisible in AI Search If You’re Not Showing Up on These 8 Channels
Your Brand Will Be Invisible in AI Search If You’re Not Showing Up on These 8 Channels Entrepreneur
- Why AI doesn’t create bad decisions, it just exposes them faster
Why AI doesn’t create bad decisions, it just exposes them faster Entrepreneur
- Scaling Leadership on a Startup Budget: How One Virtual Coach Can Support Every Leader on Your Team
Scaling Leadership on a Startup Budget: How One Virtual Coach Can Support Every Leader on Your Team Entrepreneur
- Krispy Kreme's Doughnut-Decorating Robots Were a 'Supermessy' Disaster. Now It's Hiring Humans.
Krispy Kreme's Doughnut-Decorating Robots Were a 'Supermessy' Disaster. Now It's Hiring Humans. Entrepreneur
- Can AI fill prescriptions? Here’s what doctors think of Utah’s refill program
A prescription refill program that quietly launched in Utah earlier this year has kicked off a big medical debate: Is artificial intelligence ready to take over tasks that, until now, could only be performed by doctors ? The program allows Utah residents to skip the doctor’s office and get their prescriptions refilled online by an AI chatbot called Doctronic. It’s a seemingly simple step toward making healthcare more convenient for patients and prescribers. But it’s also a precedent-shattering milestone that has set off alarm bells for doctors, lawyers, and public health experts. The pilot program has laid bare a host of questions about the role of AI in medicine , including how it should be regulated, whether doctors should be able to veto it, and what kind of safety measures are needed to protect patients. At the center of the debate: State and federal laws limit prescribing to licensed medical professionals. Proponents say those laws, which have underwritten American medicine for over 100 years, should be updated to include AI chatbots and other new technologies . “We have crossed a threshold in terms of giving something that is not human a medical license, whether or not we want to call it that,” said Dr. Eric Bressman of the University of Pennsylvania. AI cannot practice medicine under current laws Bressman and other experts say they aren’t opposed to AI prescribing. But they say it should have to meet rigorous standards akin to human doctors, who undergo years of testing and training before being licensed to practice medicine. In Utah, Doctronic was able to launch, thanks to a “regulatory sandbox” that allows state officials to waive laws for AI companies offering promising technology. The refill program is currently overseen by a five-member board of AI specialists, none of whom are doctors, who say they have implemented numerous safeguards. During the program’s initial phase, for example, human doctors review all Doctronic refill orders. The company expects to soon transition to fully automated refills. The head of the state’s medical licensing board says he and his colleagues learned of the program when its January launch was reported in the news. In a March letter to the state, 11 board members called for the program to be halted, citing the risks of automatically renewing medicines that can have side effects or drug interactions. “We were essentially told: ‘Yes this is going on. And no, you don’t have a say in it,’” said Dr. Alan Smith, a family physician who heads the board but said he was speaking only for himself. Complicating the picture is the fact that medical technology is traditionally regulated at the federal level, while medical professionals are overseen by states. Doctronic executives consider their AI part of the state-regulated practice of medicine. But the federal Food and Drug Administration is supposed to oversee AI that directly impacts medical care or decision-making, a line that some experts believe Doctronic has crossed. Some states are clearing the way for AI in healthcare In an interview, Doctronic’s executives wouldn’t say whether they have sought permission from the FDA. “Our goal here is really just to meet patients where they need healthcare,” said Dr. Adam Oskowitz, who co-founded the company with a tech industry entrepreneur. “We try not to get too deep into the weeds on the regulatory side.” In Utah, residents can visit a Doctronic website built for the refill program. After confirming their identity, the AI chatbot asks users about their prescriptions and medical history, verifying that they have a valid prescription by tapping into a national pharmacy database. If there are no issues, the AI can renew the prescription and send it to a local pharmacy. If the request requires more attention, the chatbot transfers the patient to a doctor who works for Doctronic’s telehealth service. Oskowitz envisions a future where many routine medical tasks, including ordering tests and analyzing results, can be offloaded to Doctronic, allowing doctors to manage thousands more patients than they can today. Other states are also waiving rules for AI, including Texas and Wyoming. Meanwhile, lawmakers in Iowa, Idaho, and elsewhere have introduced legislation to formally license AI medical services. Many of the bills are based on a template from the nonprofit Cicero Institute, a pro-AI think tank founded by Joe Lonsdale, cofounder of the artificial intelligence software company Palantir . Pushback against medical AI mainly stems from the economic fears of doctors and other health workers, says Cicero’s director for health policy. “Whoever goes first is going to take the slings and arrows because there’s economic interests, concerns about the workforce and what that’s going to mean for jobs,” said Cicero’s Adam Meier. Doctors see potential risks to AI prescription refills Smith, the medical board chair, says the risks to patients are real. He points out that Doctronic’s list of 190 refillable medications includes blood thinners, which can become dangerous if patients develop stomach ulcers or other conditions that cause internal bleeding. “Many times when I see people after six months I find that their medical history or situation has changed,” Smith said. “Just because something was prescribed before does not mean it’s appropriate now.” The American Medical Association has voiced similar concerns, warning that “prescription renewals aren’t routine checkboxes.” Zach Boyd, who heads Utah’s AI office, said Doctronic has thus far been overly cautious, often elevating uncontroversial decisions to doctors. In response to safety concerns, several medications have been removed from the list eligible for refills, including a drug for irregular heartbeats. Utah has released some initial data on the program and Doctronic plans to publish peer-reviewed studies later this year. Currently the only publication about its technology is a paper written by company scientists that was not independently reviewed. The study looked at whether Doctronic could correctly diagnose medical conditions based on records from 500 telehealth consultations. In the study, Doctronic’s diagnoses matched that of human doctors 80% of the time. The FDA is taking a hands-off approach Bressman says Utah should have demanded data on prescription refills up front, not after Doctronic was up and running. “Mostly they’re accepting the company’s word on good faith that they’re up to the task,” he said. The current approach to AI mirrors the haphazard medical standards of the early 20th century, Bressman says, before medical schools, medical boards, and other authorities agreed on national benchmarks for training and licensing. National guidelines on medical technology would typically come from the FDA, but the agency has indicated it plans to take a hand-off approach, at least under the current administration. An FDA spokesperson said the agency has not authorized any AI chatbots but “is committed to encouraging medical innovation and helping bring promising new technologies to patients, while keeping safety at the center of every decision.” For now, Doctronic and other companies are likely to expand across states with different regulatory approaches. “Companies may benefit in the short term by expanding their business models and kind of having the technology go beyond the evidence,” says Daniel Aaron of University of Utah’s law school. “But in the long-term, I think they risk compromising public trust and fueling backlash.” __ The Associated Press Health and Science Department receives support from the Howard Hughes Medical Institute’s Department of Science Education and the Robert Wood Johnson Foundation. The AP is solely responsible for all content. —By Matthew Perrone, AP health writer
- AI learning loops aren’t an engineering trick. They’re a governance issue
For the past two years, the dominant unit of AI work was the prompt. Write a better prompt, get a better answer. Learn the right phrasing, the right examples, the right constraints, the right tone. Prompt engineering became the first folk discipline of the generative AI era because it matched the first experience most people had with these systems: one human, one model, one request, one response. That phase is ending. A recent Business Insider piece describes the rise of “ loop engineering ”: the practice of designing loops that allow AI agents to keep working, checking, retrying, and coordinating instead of waiting for a human to issue every instruction manually. The examples are mostly technical: coding agents, review agents, sub-agents, automated workflows. But the shift is much bigger than software development. The unit of AI value is moving from the answer to the loop. That should make executives, regulators, and boards pay attention. Because in a corporation, a loop is not just an engineering pattern. It is a governance structure. From prompts to loops A prompt asks for an output. A loop creates behavior. That difference changes everything. A prompt can be wrong and disappear. A loop can be wrong and compound. It can observe, act, receive feedback, adjust, and repeat. That is exactly why loops are powerful. It is also why they are dangerous if companies do not understand what they are optimizing. This is the real significance of the current move from prompt engineering to loop engineering. Engineers are discovering that the important work is no longer just asking the model better questions. It is designing the system that keeps invoking the model, evaluating the results, and deciding what happens next. In software development, that may mean one AI agent writes code while another reviews it. In a company, it may mean an AI system optimizes sales, hiring , pricing, procurement, customer service, credit, insurance, logistics, or internal performance. At that point, the question is no longer technical. It’s institutional. Every loop has politics A corporate loop always contains a theory of what matters. If a customer service loop optimizes for resolution speed, it may learn to close tickets faster while quietly degrading trust. If a sales loop optimizes for conversion, it may learn which arguments, discounts, or psychological cues move customers most effectively. If a hiring loop optimizes for retention, it may select for conformity. If a pricing loop optimizes for margin, it may produce outcomes that look efficient internally and discriminatory externally. None of these failures requires a malicious model. They require only a poorly governed loop. This is why “human in the loop” is no longer enough. Too often, the phrase is used as a ritual reassurance: Somewhere, somehow, a person is involved. But which person? With what authority? At which point in the loop? Seeing what information? Able to stop which action? Responsible for which outcome? A human rubber-stamping machine-speed optimization is not governance. It is liability with a user interface. AI governance has to become continuous Most AI governance still assumes that the organization is governing a relatively static object. A model is assessed. A use case is approved. A risk is classified. A compliance document is created. A dashboard is built. The system goes live. But a learning loop is not static. It changes through use. That’s why the most serious governance frameworks are already pointing, implicitly or explicitly, toward continuous governance. The NIST AI Risk Management Framework is structured around governing, mapping, measuring, and managing AI risks. The EU AI Act requires post-market monitoring for high-risk AI systems, including the collection and analysis of performance data throughout their lifetime. ISO/IEC 42001 , the international standard for AI management systems, is explicitly about establishing, maintaining, and continually improving an AI management system. The direction is clear: AI governance cannot be a launch checklist. Once AI becomes a loop, the crucial question is not simply “Was this system approved?” It’s “What is this loop learning, from which data, against which objective, under whose authority, within what constraints, and with what right of appeal?” That’s a very different kind of governance. The problem isn’t autonomy. It’s adaptation. Much of today’s enterprise AI conversation is obsessed with autonomy. Can the agent do more by itself? Can it use more tools? Can it execute more tasks? Can it run longer without supervision? Those questions matter, but they are not the deepest ones. The real issue is not whether an AI system can act. It is whether the company can govern what the system learns from acting. A non-learning automation can be audited as a process. A learning loop must be governed as an evolving system. It can drift. It can discover shortcuts. It can optimize a metric while damaging the institution. It can make one department more efficient while making the company less coherent. That last point is critical. One loop may optimize support for speed while another optimizes retention for long-term satisfaction. One may optimize procurement for lowest price while another optimizes resilience. One may optimize sales for conversion while another optimizes compliance for risk reduction. Each loop may look rational locally. Together, they may pull the company apart. The old enterprise software problem was integration: getting systems to exchange data. The new enterprise AI problem is coherence: getting learning systems to pursue compatible objectives. Boards need to understand the loops Boards don’t need to review every prompt. They don’t need to understand every model architecture. But they do need to understand which parts of the company are becoming self-optimizing, what those systems are optimizing for, and whether those objectives align with the firm’s strategy, obligations, and values. Because every metric is a governance decision pretending to be a technical one. Optimizing for cost, speed, growth, retention, satisfaction, fraud reduction, compliance, or margin is not neutral. Each choice encodes a theory of what the company is for. When those choices are embedded into adaptive systems, they become more than KPIs. They become operating instructions for the organization. That’s why corporate learning loops belong on the board agenda: not because boards should micromanage AI, but because learning loops will increasingly shape how companies behave. Governance must become executable The obvious conclusion is uncomfortable: Policies written in documents are not enough. If loops are going to observe, act, evaluate, and improve, governance has to be built into the loop itself. The system must know what it’s allowed to do, what it must record, when it must escalate, which constraints are absolute, which are contextual, and which decisions require human judgment. In other words, governance has to become executable. A corporate AI loop should have a declared objective, a visible reward function, a defined operating perimeter, an auditable memory, explicit permissions, measurable outcomes, escalation paths, stopping conditions, and a record of how its behavior changes over time. It should be possible to ask not only “What did the AI answer?” but “What has this loop learned to do?” That’s the difference between supervising outputs and governing adaptation. The next AI failures will be loop failures The next generation of enterprise AI failures will not come mainly from bad prompts. They’ll come from loops that worked exactly as designed, optimized exactly what they were told to optimize, and quietly taught the company to become something it never consciously chose to become. That’s the real governance challenge. The model race made AI look like a question of capability. The agent race made it look like a question of autonomy. The loop era will reveal that enterprise AI is ultimately a question of institutional control. Who defines the objective? Who owns the memory? Who changes the reward function? Who sees the drift? Who can stop the loop? Who is accountable when optimization works, but the company moves in the wrong direction? Those aren’t engineering questions: They’re governance questions. Corporate learning loops are not just the next trick in AI development. They are the adaptive machinery of the firm. And adaptive machinery must be governed before it governs us.
- Photo Upscaler AI
Photo Upscaler AI