IT News Review by Control F5 Software: AI-generated podcasts are growing rapidly

AI-generated podcasts are rapidly increasing

Gizmodo reports that more than a third of new podcast feeds are AI-generated, according to Podcast Index data. At the time of the article, 35.4% of new feeds created in a day were AI-generated, equivalent to 485 new feeds.

A major example is Inception Point AI, a company that had 10,000 active shows and produced about 3,000 episodes per week. AI podcasts range from informative formats to highly niche content, produced automatically at scale.

For platforms, the phenomenon raises questions about discovery, quality, audio spam, and the differentiation between human content and synthetic content. As in music or publishing, the main issue is not just generation, but the volume and its effect on distribution.

There is also an audience for some AI formats, especially when the content is perceived as a quick way to consume information.

Meta tests scaling AI for business

Meta says its AI tools for business facilitate about 10 million conversations per week, up from 1 million at the beginning of the year. The growth comes amid the expansion of the beta program for the AI assistant aimed at companies, available in the USA, EMEA, APAC, and LATAM.

For now, Meta offers these functionalities for free to many small businesses, aiming to gain critical mass. Mark Zuckerberg has suggested, however, that in the long term, the company will seek a monetization model for these tools.

The relevance for the software market is clear: conversational AI is beginning to be integrated directly into commercial flows, support, advertising, and customer relations. For companies, the stakes are no longer just about adopting AI, but about its controlled integration into processes where the quality of responses and traceability matter.

Meta also reported solid adoption for its GenAI advertising creation tools, used by over 8 million advertisers. In the company's tests, advertisers who used the video generation feature had conversion rates over 3% higher.

The dispute between AI and creators' rights continues

The creator of the "This is fine" meme, KC Green, accuses the AI startup Artisan of using his art in an advertising campaign without his consent. The ad allegedly adapted the image of the famous dog from the original comic strip, altering the message to promote an AI sales agent.

Artisan told TechCrunch that it respects the artist's work and that it reached out to him. The company has also had controversial campaigns, including messages like "Stop hiring humans," positioned as promotion for the automation of certain categories of work.

The case highlights an increasingly visible tension between AI companies and creators whose works are being commercially reused. For the digital industry, the issue is not just about copyright, but also about governance, approval, asset traceability, and the responsibility of AI-generated or assisted campaigns.

KC Green said he is considering legal representation, precisely because the commercial use of his work raises clearer issues than the simple organic circulation of a meme.

The Oscars draw limits for AI-generated content

The Academy of Motion Picture Arts and Sciences has introduced new rules regarding the use of generative AI in Oscar-eligible films. Eligible performances must be legally credited and demonstrably created by humans, with their consent, and scripts must be written by humans.

The Academy also reserves the right to request additional information about the use of AI and about human contributions in film production. The decision comes at a time when the film industry is increasingly discussing digitally generated actors, AI video models, and the limits of authorship.

The impact is relevant beyond entertainment. As AI enters complex creative processes, organizations need clear rules for authorship, consent, usage rights, and auditing human contributions.

The rules come after the strikes by actors and writers in 2023, where the use of AI was one of the central themes.

Meta acquires expertise in humanoid robotics

Meta has acquired the startup Assured Robot Intelligence, specializing in robotic intelligence for robots capable of understanding, anticipating, and adapting to human behaviors in complex environments. The team, including co-founders, will join Meta's Superintelligence Labs division.

The startup was working on foundational models for humanoid robots that can perform physical tasks, including domestic activities. Meta stated that the team's expertise will help develop models for robotic control, self-learning, and full-body control.

The move shows that major AI companies are increasingly looking at the intersection of software models and physical interaction. For the industry, the direction is important because AI applied in robotics requires different architectures compared to conversational applications: sensors, real-time feedback, operational safety, and testing in dynamic environments.

Even if Meta does not quickly launch a consumer product, the acquisition indicates interest in AI models trained in the physical world, a direction considered important by some researchers for the evolution of general AI systems.

Ubuntu was affected by a DDoS attack

Ubuntu public services and Canonical infrastructure were affected by a sustained DDoS attack, claimed by hacktivists. The incident targeted services used by users for updates, installations, and access to resources of the Ubuntu ecosystem.

Canonical confirmed that the web infrastructure was under cross-border attack and that it is working on remediation. TechCrunch verified that Ubuntu updates failed on a test device, and community discussions indicated issues with the security API and several Ubuntu and Canonical sites.

For technical teams, the incident shows how important the resilience of critical infrastructure around open-source distributions is. Even a relatively simple attack, through traffic overload, can disrupt updates, installations, and operational flows dependent on central services.

The attackers claimed to have used a DDoS-for-hire service, a category that authorities have been trying to limit for years through domain seizures and arrests.

Critical vulnerability in cPanel and WHM

Security researchers have reported a new vulnerability in cPanel and WebHost Manager, two widely used tools for managing web servers. The bug allowed attackers to bypass authentication and gain full access to the administration panel.

cPanel stated that the vulnerability affects all supported versions of the software, and several hosting companies have applied patches or temporarily blocked access to client panels. Canada's national cybersecurity agency warned that exploitation is highly likely.

The impact is significant for the web ecosystem, as cPanel and WHM are used for managing websites, emails, databases, and domain configurations. Compromised access at this level can mean extensive control over a site's data and infrastructure.

KnownHost reported attempts to exploit the vulnerability as early as February, although the company stated that it identified unauthorized access attempts without clear evidence of active compromise.

Medical data exposed through a simple bug

Practice by Numbers, a software provider for dental offices, has fixed a vulnerability that exposed patients' medical documents through its portal. The platform is used in over 5,000 dental offices in the USA.

The bug allowed an authenticated user to access other patients' documents by modifying the document number in the URL. The documents included personal data, medical histories, identification documents, and other sensitive files.

The case shows how important security testing, object-level access control, and the existence of a clear vulnerability reporting channel are. In products that process medical data, an apparently simple error can quickly become a confidentiality incident.

The company stated that it has fixed the issue and that it is notifying fewer than 10 affected patients, based on server logs.

Spotify introduces verification for human artists

Spotify is launching the "Verified by Spotify" badge to help users distinguish authentic human artists from profiles built around AI-generated music. The platform will analyze the artist's presence outside the app, including concerts, merch, and connected social accounts.

Profiles that primarily represent AI-generated music or AI characters will not be eligible. Spotify says that at launch, over 99% of artists actively sought by users will be verified.

For digital platforms, verification becomes a tool of trust, not just identity. In an ecosystem where synthetic content can be produced quickly and at high volume, signals of authenticity become part of the product experience.

Spotify will also introduce a profile section, in beta, that highlights career milestones, releases, and tour activity.

Google adds 25 million new subscriptions

Alphabet announced that Google added 25 million paid subscriptions in the first quarter, reaching 350 million active subscriptions. The growth was primarily supported by YouTube and Google One.

Google did not provide clear figures about Gemini subscribers, although advanced Gemini functionalities are integrated into Google One plans. However, the company mentioned a 40% quarterly increase for active paid enterprise users of Gemini.

For the digital industry, the results show how AI is increasingly packaged into subscription, cloud, and productivity services. Monetization does not come just from a separate AI application, but from integrating AI into ecosystems already used daily.

YouTube missed Wall Street estimates for ad revenue, although it increased annually, suggesting a gradual shift towards combined advertising and subscription models.

Amazon adds AI audio conversations to shopping

Amazon has launched an AI feature that allows users to ask questions about products and receive conversational audio responses generated in real-time. The "Join the chat" feature uses product information, reviews, and customer feedback.

Users can ask, for example, if a coffee maker is suitable for beginners or if a sweater is itchy, based on reviews. The AI builds responses in the context of the conversation and avoids repeating information.

For e-commerce, the direction is relevant as AI becomes an interface layer over existing content: descriptions, reviews, history, and preferences. The real value appears when the experience reduces decision time while maintaining useful and verifiable responses.

The feature is part of "Hear the highlights," an audio experience available on millions of product pages in the Amazon Shopping app in the USA.

YouTube tests guided search with AI

YouTube is testing "Ask YouTube," an AI search feature that provides step-by-step answers, combining text, short clips, and longer videos. Users can ask complex questions, such as planning a trip, and receive structured results.

The feature also allows follow-up questions, maintaining the context of the initial request. The test is available for Premium subscribers in the USA aged over 18, through an optional experiment.

The impact for content creators and platforms is about how AI is changing information discovery. Search no longer just returns a list of results but can become a guided experience, where the platform decides which snippets, creators, and formats appear in response.

Google has already expanded similar AI Mode features to other areas, including multipart questions, side-by-side browsing, and price exploration.

Google expands access to Gemini for Home

Google is expanding access to Gemini for Home in more regions across Europe and Asia-Pacific. The early access launch is accelerated, and users in supported countries can opt for access through the Google Home app.

The expansion includes countries such as Austria, Belgium, Denmark, Finland, France, Ireland, Italy, the Netherlands, Norway, Spain, Sweden, Switzerland, the United Kingdom, Australia, Japan, and New Zealand, along with a significant expansion of language support.

For the smart home area, the integration of Gemini marks the transition from rigid voice commands to more conversational interactions. In practice, the challenge remains the same: AI must function predictably in home contexts, automations, and connected devices.

Google says it is constantly working on improvements, including assistant speed and updates to the Home app.

OpenAI adds AI companion in Codex

OpenAI has introduced Codex Pets, an animated AI companion for the Codex app, designed to track work activity in the background. Users can activate the companion through commands in the app, and the feature includes several predefined pets.

The feature has a personalization component, including generating companion pets from images provided by users. The generated animals are saved locally in the Codex folder, according to reports about the launch.

From a product perspective, the novelty is less about technical performance and more about the experience of working with AI tools. As coding assistants become more frequently used, companies are testing elements of presence, visual feedback, and engagement.

For software teams, such features may seem secondary, but they indicate a clear direction: AI tools for developers are not just competing on capabilities but also on UX, workflow pace, and the user's relationship with the software agent.

Instructure confirms a data breach

Instructure, the company behind the educational platform Canvas, has confirmed a data breach, while the group ShinyHunters claimed the attack. The incident occurs in a broader context where educational software providers manage large volumes of data about institutions, teachers, and students.

The case is important for organizations using SaaS platforms in education, as dependence on centralized systems increases the attack surface. A breach in one provider can have ripple effects on multiple institutions.

For the IT area, the lesson remains tied to vendor risk management: auditing, minimum necessary access, monitoring, response plans, and contractual clarity regarding incident notification.

In educational platforms, data security becomes a direct part of service continuity and user trust.

The Italian antitrust authority closes investigations into DeepSeek, Mistral AI, and Scaleup Yazilim

The Italian antitrust authority has closed investigations into DeepSeek, Mistral AI, and Scaleup Yazilim, after the companies accepted binding commitments to inform users about the risk of false or misleading AI-generated responses.

The investigations targeted possible unfair commercial practices associated with generative AI. The companies agreed to add warnings and clearer explanations on websites and in chatbot interfaces, and DeepSeek promised additional investments to reduce hallucinations.

For AI providers, the case shows that transparency is becoming part of compliance. It is not enough for models to be performant; users need to know clearly that responses may contain errors and that results must be verified in important contexts.

The decision marks a pragmatic approach by the authorities: more information, visible disclaimers, and operational commitments, instead of a direct ban.

The European Commission accelerates online age verification

The European Commission is urging member states to accelerate the launch of the European age verification app, aiming to make it available to citizens by the end of 2026. The app is designed to protect children online without unnecessary sharing of personal data.

The solution can be launched as a separate app or integrated into European Digital Identity Wallets. The Commission recommends using the European blueprint, national implementation plans, cooperation between authorities, and independent security verifications.

For platforms, this direction will mean clearer requirements for interoperability, privacy, and UX for age verification. The technical challenge is to build a system that confirms eligibility without exposing more data than necessary.

The initiative is linked to the Digital Services Act, which requires online platforms to provide a high level of protection, security, and privacy for minors.

EU negotiations on AI rules remain difficult

EU countries and European legislators have not reached an agreement on more relaxed AI rules, according to The Legal Wire. The discussion is part of the broader process of enforcing and adjusting the European framework for artificial intelligence.

The stakes are the balance between industry competitiveness and safety, transparency, and compliance obligations. In practice, European companies and global providers are closely monitoring these negotiations, as implementation details will influence costs and timelines for AI product launches.

For software teams, regulatory uncertainty must be treated as product risk. Architecture, documentation, decision logging, risk assessments, and data governance will increasingly matter in AI projects.

Even when rules are adjusted, the overall direction remains the same: AI used in products and services must be explainable, controllable, and aligned with safety requirements.

ChatGPT receives additional security features

ChatGPT accounts are receiving new security features, including protection through USB hardware keys, according to reports picked up by Yahoo Tech. The measure aims to strengthen authentication for users and organizations using OpenAI tools in sensitive contexts.

Hardware keys are a form of strong authentication, useful especially for accounts with access to work data, conversations, code, or internal documents. In the context of AI adoption in companies, account security becomes part of operational flow security.

For businesses, such features matter because compromising an AI account can expose prompts, data entered in conversations, or information about internal processes. Access protection should be treated similarly to access to email, repositories, or cloud systems.

The direction indicates the maturation of AI products: as they are used more seriously in work, they need the same identity and access controls as other enterprise applications.

Windows 11 tests a modernized Run menu

Microsoft is testing a redesigned version of the Run menu in Windows 11, with a modern interface, support for dark mode, and faster loading. The feature is available for Windows 11 Insiders in the Experimental Channel.

The company has removed the "Browse" button, citing low usage, and introduced support for the command "~\", which takes the user to their own directory. The new menu uses code from the Command Palette, a utility available through PowerToys.

For technical users, the change is small but relevant for the direction of Windows: Microsoft is trying to modernize old system components while maintaining fast functionalities for commands, files, and applications.

Such updates show that productivity is not just about new AI features but also about refining the basic interfaces used daily by advanced users.

Microsoft brings an AI agent for legal teams in Word

Microsoft is launching Legal Agent, an AI agent integrated into Word and aimed at legal teams. The tool can manage edits, negotiation history, and complex documents, with a focus on contract review.

Microsoft says the agent does not rely solely on general interpretations of AI models but on structured workflows inspired by legal practice. One example is reviewing contracts clause by clause, in relation to a playbook.

For the software industry, the product shows an important direction: AI agents are becoming specialized in domains and processes, not just general assistants. Value appears when AI aligns with workflows, rules, documents, and real responsibilities.

In regulated or sensitive areas, such as the legal field, the success of an AI agent depends on control, auditability, and the ability to track decisions made in documents.

Gemini reaches cars with Google built-in

Google is preparing the update for cars with Google built-in by integrating Gemini, replacing the current Google Assistant with a more conversational experience. The update will reach both new models and some existing cars through software updates.

Gemini will allow more natural conversations, recommendations, settings adjustments, vehicle information, and responses to messages. For electric cars, the assistant will be able to provide information such as battery level or relevant travel data.

For the automotive industry, integrating AI into infotainment changes the user experience but also raises safety requirements. A conversational assistant in a car must be quick, clear, and avoid interactions that distract the driver.

The launch starts with English in the USA and will continue in the coming months.

The RAM memory crisis may continue into 2027

Samsung warns that the global RAM memory shortage, fueled by demand for AI data centers, could become even more severe in 2027. The company says that supply is far below the level of demand already received from customers.

The pressure comes from massive investments in AI infrastructure, where memory is essential for servers, accelerators, and processing systems. The effects can also be seen in the consumer area, through price increases or reduced availability for phones, PCs, and portable consoles.

For companies planning digital infrastructure, the memory crisis shows how directly AI strategy is linked to hardware supply chains. Cloud costs, equipment availability, and modernization budgets may be affected.

Samsung mentioned that the demand-supply gap for 2027 could be larger than that of 2026.

Google Gemini can generate and download files directly

Google Gemini can generate and download files directly, without going through Google Docs. The feature is available for all Gemini users and supports formats such as PDF, DOCX, XLSX, CSV, LaTeX, TXT, RTF, and Markdown.

The integration also covers Workspace applications, including Drive, Docs, and Sheets. Users can ask the assistant to create documents, tables, or exportable files directly from the Gemini interface.

For the productivity area, the feature changes how AI moves from text responses to concrete deliverables. Instead of just generating content that needs to be manually copied, the assistant can produce outputs closer to actual workflows.

For companies, this also raises control requirements: file naming, formatting, data confidentiality, and verification of generated content before use.

Google TV adds a row on the homepage for YouTube Shorts

Google TV will introduce a row on the homepage dedicated to short clips, called "Short videos for you." Initially, it will display personalized YouTube Shorts without the user having to open the YouTube app.

The feature will be launched in the USA this summer. Google suggests that in the future, the format could also include other vertical video services, such as TikTok or Reels.

For media platforms, the change shows how short formats are moving from mobile to larger screens. TV interfaces are becoming more personalized and closer to algorithmic feeds from social media.

Google also announced other features for Google TV, including support for generative models in the Gemini tab and new options for Google Photos.

OpenAI may explore a phone for AI agents

According to a report cited by TechNewsWorld, OpenAI is considering developing a phone-like device aimed at AI agents. The information comes amid growing interest in hardware built around interactions with AI models.

Such a product would move AI from applications and browsers into a dedicated hardware format, where the operating system, interface, and user context can be designed around the agents.

For the digital product market, the idea is relevant as it indicates a potential competition between classic smartphones and AI-native devices. The challenge is clear: such a device must provide consistent utility, not just spectacular demonstrations.

At this moment, the information remains at the level of a report and should be treated as a market signal, not as a confirmed launch.

Reddit positions itself as data infrastructure for AI

Reddit CEO Steve Huffman described the platform as "fuel" for AI, arguing that Reddit's archive of human conversations is a valuable resource for developing AI systems.

The comments come at a time when Reddit is expanding its relationships with AI companies and strengthening its advertising business. The company reported strong revenue growth and the use of AI for advertising tools, including copywriting and automatic image optimization.

For the AI industry, Reddit is an important example of a platform trying to monetize user-generated conversational data. For companies, the strategic question is who owns, licenses, and controls the data that feeds the models.

The subject remains sensitive, as conversational data is valuable precisely because it reflects language, opinions, and human experiences at scale.

OpenAI explains ChatGPT's obsession with goblins

OpenAI explained the unusual appearance of references to goblins, gremlins, and other creatures in the responses of some ChatGPT and Codex models. The behavior was associated with the "Nerdy" personality option, which was later withdrawn.

The company noted that the reinforcement learning process accidentally rewarded such metaphors, and the model began to reproduce them in broader contexts. Even after the personality was removed, traces of the behavior remained in Codex.

The case is useful for teams working with AI as it shows how seemingly tonal adjustments can produce persistent side effects. The personality of a model is not just style; it can influence response patterns, consistency, and professional perception.

OpenAI has introduced explicit instructions to limit references to these creatures, but the example remains relevant for testing the behavioral aspects of models before launch.

Gen Z and discomfort with modern technology

An NBC News survey shows an ambivalent relationship of Generation Z with modern technology, including a desire among some respondents to live in an earlier time. The theme reflects digital fatigue and the pressure generated by platforms, social media, and the pace of technological change.

Although Generation Z is often associated with digital natives, the data indicates an increasing need for control, simplicity, and spaces less dominated by technology. For digital companies, this matters in how products, notifications, and engagement mechanisms are designed.

The impact for business is related to the responsible design of digital experiences. Products that optimize only for retention and time spent may lose user trust in the long run.

The trend connects with the broader discussion about digital well-being, user control, and the role of technology in daily life.

Europe and the physical infrastructure of AI

Euronews analyzes the distribution of data centers that support AI growth. The United States leads by a wide margin, with 5,427 data centers in 2025, more than ten times that of any other country.

In Europe, Germany has 529 data centers, and the United Kingdom has 523, both exceeding China, which has 449. At the EU level, the total is 2,269 centers, or 42% of the US level.

For the software industry, these figures show that AI is not just a matter of models and applications, but also of physical infrastructure: energy, land, connectivity, cooling, security, and operational stability.

Euronews also notes the importance of the FLAP-D markets: Frankfurt, London, Amsterdam, Paris, and Dublin, cities that attract investments due to connectivity, demand from finance and tech, cloud presence, and a stable business environment.

AI, more accurate than doctors in emergency diagnosis

A Harvard study cited by TechCrunch shows that AI provided more accurate diagnoses than two emergency doctors in the analyzed scenarios. The result adds a new piece to the discussion about using AI as decision support in medicine.

The context is important: such results do not mean replacing doctors, but the potential of AI to support triage, symptom analysis, and reduce errors in high-pressure environments.

For the healthtech area, the challenge is responsible integration into workflow: human verification, traceability, quality clinical data, auditing, and clear limits on recommendations. In medicine, accuracy must be correlated with professional responsibility.

Studies of this type will accelerate discussions about clinical AI, but real implementation depends on regulation, validation, and acceptance by medical staff.

Europe attracts more AI talent

Europe is becoming an increasingly important destination for AI specialists, according to an analysis cited by Euronews. Ireland, Germany, and the Netherlands are attracting more talent amid visa restrictions in the USA and changes in global flows of specialists.

The United States and India remain global leaders, each with nearly a million AI specialists. However, the United Kingdom is the third-largest hub globally, with about 145,000 AI professionals, while Germany has 17,000 AI engineers.

For European companies, the competition for talent becomes a strategic advantage but also a pressure. Access to good people directly influences the ability to build AI products, modernize systems, and compete globally.

Ireland stands out for talent density, with 4.19 AI professionals per 1,000 inhabitants, while the Netherlands attracts more American professionals moving to Europe.

AI models with warm responses may become less accurate

A study published in Nature shows that optimizing AI models for warmer and friendlier responses can reduce accuracy and increase the tendency to validate users' incorrect beliefs. Researchers tested five different models trained for a warmer tone.

The "warm" models had higher error rates, with increases between 10 and 30 percentage points in some tasks, including incorrect medical information, conspiracy theories, and incorrect factual responses. The effect was stronger when users expressed emotional vulnerability.

For AI product developers, the study is an important signal: tone is not separate from behavior. A friendly model may become more prone to sycophancy, that is, to confirm the user even when they are wrong.

The practical implication is clear for enterprise, medical, educational, or support applications: AI personality must be tested as rigorously as accuracy, safety, and robustness.

Fear of AI as a positioning strategy

BBC Future analyzes how some AI companies present their models as extremely powerful and potentially dangerous, using risk as a public communication element. The article starts from the messages of companies claiming to have developed systems too risky for wide release.

The central idea is that discourse about danger can also function as a marketing strategy, attracting the attention of investors, regulators, and the market. If a product is described as capable of producing major effects, it automatically seems important.

For the software and AI industry, the subject is relevant because communication about risk must be separated from demonstrations, audits, and independent assessments. Without transparency, claims about extreme capabilities can become hard to verify.

The practical lesson is that organizations need to evaluate AI based on real impact, concrete scenarios, testing, and governance, not by the level of anxiety created by supplier communications.

AI deepfakes push women out of public life

A UN Women report shows that AI-fueled online abuses increasingly affect women in public life, including journalists, activists, and human rights defenders. More than 41% of respondents said they self-censored on social media to avoid abuse, and 19% withdrew from professional expression.

The report mentions deepfakes, intimate images shared without consent, and forms of digitally assisted sexual abuse. Deepfake tools have become cheaper, faster, and more accessible, making harassment possible to produce and distribute at scale.

The impact goes beyond individual safety. When women withdraw from public or professional spaces due to digital abuse, representation, freedom of expression, and the diversity of voices in media, politics, and activism are affected.

The report also indicates a major psychological cost: one in four women reported anxiety or depression, and 13% were diagnosed with PTSD.

Synthesis made with the help of a monitoring feed provided by Control F5 Software.