Operations | Monitoring | ITSM | DevOps | Cloud

In a recent survey, Udacity’s team explored the evolving landscape of AI adoption by asking 2000 professionals (including those in the legal sector) if they used AI. Unsurprisingly, over 90% of respondents said they did. More concerning, 72% of managers reported personally paying out of pocket for AI tools to use at work, introducing uncontrolled risk into corporate environments.

It’s been a noisy summer in the AI world. Headlines have been filled with doom and gloom: For example, OpenAI’s ChatGPT-5 landing with a thud, and an MIT report claiming 95% of AI pilots are failing. For the sceptics, this is “proof” that AI is just hype. I don’t buy it. The MIT study looked at just 50 projects, a sample size so small you’d fail a basic stats exam for using it. And as someone who uses AI every single day, I can tell you the benefits are real.

The release of OpenAI’s new open-source model, GPT-OOS, has sparked a wave of excitement across the AI community. And rightly so. For organizations that want the benefits of generative AI without sending data out to the web, this is a compelling option. Running locally, GPT-OOS offers a level of privacy, control, and cost-efficiency that’s hard to ignore. It’s fast, lean and at least in its early benchmarks, surprisingly capable in coding, math, and STEM-heavy workloads.

August 5, 2025, Las Vegas Black Hat 2025, Abnormal AI officially launched its Security Posture Management for Microsoft 365. This release marks a critical turning point. In an era where attackers weaponized AI to uncover and exploit misconfigured cloud environments at machine speed, reactive security simply can’t keep pace. Threat actors are now leveraging automated AI to scan systems, identify configuration drift, escalate privileges, and deploy zero‑day exploits in seconds.

If, like me, you’re continuously striving to keep pace with the ever-evolving world of artificial intelligence, you’re probably hearing a lot about how Ambient AI is poised to dominate discussions and developments throughout the second half of 2025. Ambient AI refers to artificial intelligence systems that operate unobtrusively in the background of our daily environments, constantly sensing, analyzing, and responding to various inputs without explicit human interaction.

Back in 2018, I wondered (perhaps loudly if SD-WAN was just IT’s hype-of-the-year, destined for the same eye-rolls as signature-based antivirus and GDPR compliance drives. Even then, I knew we couldn’t let messaging fatigue blind us to real technology shifts. Fast-forward to 2025: SD-WAN (Software-Defined Wide Area Network) not only stuck around, but became the springboard to something far bigger – SASE (Secure Access Service Edge).

A recent BBC article, “Weak password allowed hackers to sink a 158-year-old company,” outlined a serious security lapse. This case reinforces the message that we, at Teneo, advocate every day: true resilience comes from defense in depth, i.e. policy, product and process, not just tools at the edge. In a recent customer engagement, we discussed a transition from VPN to ZTNA. While ZTNA offers enhanced security including continual checking, improved segmentation and a minimized attack surface.

Yesterday, like many others in the tech community, I found myself pausing to fully grasp the implications of the Microsoft SharePoint hack. As one of the most widely adopted document management and collaboration platforms globally, SharePoint’s compromise inevitably sends ripples of concern through businesses everywhere. This news reminded me of a conversation I had just last week with an enterprise customer. We were discussing how one might approach cybersecurity from a hacker’s perspective.

The recent launch of Grok 4 has set the AI community buzzing. With an impressive score of 73 on TLDR’s AI benchmark, Grok 4 edges ahead of OpenAI’s O3 and Google’s Gemini 2.5 Pro, both scoring 70. Elon and the X AI team deserve praise for this breakthrough, reinforcing Grok 4’s status as potentially the most powerful LLM yet.

It’s only a matter of time. OpenAI CEO Sam Altman said in 2024 that he thought this could be achieved by the end of 2026. Personally, I feel this is a little optimistic; however, based on the evidence I’ve seen, it won’t be long after that. Consider Telegram: a global messaging giant with just 30 employees, already achieving a remarkable $1 billion in revenue. Or Midjourney, revolutionizing creative industries with only 40 employees and generating an impressive $500 million.