News observatory

News

Observatory of world news on information technology, cybersecurity, resilience and continuity, and artificial intelligence. Each item links to the outlet that published it; the summaries are our own and the full content belongs to each source.

Last updated: . Summaries are our own; original headlines and full content belong to the cited media.

Cybersecurity ·

Apple patches about 200 vulnerabilities with iOS 27, iPadOS 27 and macOS Golden Gate 27

The releases published on 15 September fix around 200 flaws, about a hundred of them shared by mobile and desktop systems, including kernel issues enabling memory corruption, privilege escalation and information leaks; a media-processing vulnerability (CVE-2026-64752) was closed by removing the affected code. Organizations with same-day update support can reach a current security posture within hours.

Source: SecurityWeek · Ionut Arghire · Apple Patches 200 Vulnerabilities With New iOS 27, macOS Golden Gate 27 Releases

Read at the source →

Cybersecurity ·

Texas utility CenterPoint Energy confirms a breach after customer data is leaked

The Houston-based utility, with about seven million customers in four states, confirmed to the SEC an incident on internet-facing systems after an attacker claimed to have stolen around 7.5 million records; electricity service was not affected. It is the third known incident involving the company's data, after those linked to the Cl0p group's MOVEit campaign in 2023.

Source: SecurityWeek · Eduard Kovacs · Texas Utility CenterPoint Energy Confirms Breach After Hacker Leaks Data

Read at the source →

Cybersecurity ·

Thai broadband provider compromised through a Fortinet vulnerability

The attackers exploited CVE-2024-21762, a FortiGate SSL-VPN vulnerability, to access operator 3BB's network; their toolkit (reconnaissance, brute force, privilege escalation and MeshCentral as command and control) was found in an exposed directory of 298 files. According to Hunt.io, they attempted to extract SSH keys, PHP configurations and database credentials, and covered their tracks while keeping backdoors.

Source: SecurityWeek · Ionut Arghire · Thai Broadband Provider Hacked via Fortinet Vulnerability

Read at the source →

Artificial intelligence ·

OpenAI, Anthropic and Google DeepMind have been coordinating AI safety measures for weeks

OpenAI's global policy chief Chris Lehane confirmed that the three companies are collaborating on safety measures for frontier models, following Anthropic CEO Dario Amodei's essay urging the industry to slow the pace of development to avoid catastrophic risks. The coordination could draw antitrust scrutiny, while the US government prioritizes competitive advantage over China.

Source: TechCrunch · Rebecca Bellan · OpenAI, Anthropic, Google have been in talks on AI safety for weeks

Read at the source →

Artificial intelligence ·

Microsoft publishes a draft code of conduct for its AI models with cyberattack boundaries

The draft "Humanist AI Code of Conduct" prevents the models from producing working exploit code, attack tooling or planning and targeting methodologies, while permitting defensive work; it establishes a chain of command that stops external content from overriding safety constraints and limits autonomous agents to the user-defined scope with minimum privileges. Microsoft opened a six-week public consultation before finalizing the guidelines for its 2027 models.

Source: SecurityWeek · Eduard Kovacs · Microsoft AI Code of Conduct Sets Cyberattack Boundaries, Chain of Command, Safety Constraints

Read at the source →

Artificial intelligence ·

Salesforce and Nvidia unveil Koa, a reasoning model specialized in enterprise tasks

Koa was built on Nvidia's open-weight Nemotron base for sales, marketing and customer support, and was trained on synthetic data rather than real customer information, reducing data-security risks and token consumption. The collaboration points to specialized enterprise alternatives that run inside each organization's infrastructure while preserving data sovereignty.

Source: TechCrunch · Julie Bort · Salesforce and Nvidia's new reasoning model is everything the AI labs should fear

Read at the source →

Resilience and continuity ·

The BCI publishes a paper on cyber resilience of transport systems

The Business Continuity Institute's cyber resilience special interest group published a member-only paper that examines emerging threats to transport networks, assesses their operational and safety implications, and proposes practical strategies to strengthen resilience. The digital transformation of transport exposes critical infrastructure to attacks capable of disrupting operations and compromising public safety.

Source: Business Continuity Institute (BCI) · Building Resilient Transport Systems Against Evolving Cyber Threats

Read at the source →

Information technology ·

Teams and Copilot are changing addresses: organizations must update their firewalls

Starting this month Microsoft is redirecting Teams and Microsoft 365 web users to teams.cloud.microsoft and copilot.cloud.microsoft. IT teams must update firewalls, proxies, gateways and network controls by early October to keep access; organizations that cannot meet the deadline can request limited exceptions for Teams, available only until 31 December 2026.

Source: Computerworld · Maxwell Cooter · Teams and Copilot are changing addresses: update your firewalls

Read at the source →

Information technology ·

Apple introduces the A20 Pro, the first 2-nanometer processor in a mass-market phone

The iPhone 18 Pro chip features a dual 16-core Neural Engine, 50 % more memory bandwidth and up to 40 % more GPU performance than the A19 Pro. Its wafer-level multi-chip module packaging places memory beside the processor instead of stacking it, improving heat dissipation; Qualcomm and MediaTek are preparing similar 2-nanometer approaches.

Source: Computerworld · Jonny Evans · Apple's A20 Pro rewrites the rules for chip design — again

Read at the source →

Information technology ·

The change of Automattic's CEO reopens the debate on WordPress governance

Automattic's board placed Matt Mullenweg on paid leave and appointed CFO Mark Davies as interim CEO. The decision concerns organizations using WordPress, present on more than 40 % of the web, because Mullenweg personally controls WordPress.org, which manages the platform's security patches and updates; according to analysts, that concentration of control is a governance risk the CEO change does not resolve.

Source: Computerworld · Evan Schuman · Automattic CEO Matt Mullenweg is out: Does this mean long-term viability, or liability, for WordPress customers?

Read at the source →

Resilience and continuity ·

Overlapping outages of four AI services expose an enterprise resilience gap

On 3 September 2026, OpenAI's ChatGPT, Anthropic's Claude, xAI's Grok and Google's Gemini suffered interruptions within hours of each other, lasting three to four hours and affecting millions of users. The analysis notes that many organizations have integrated AI into customer service, development, security and analytics without redundancy plans, turning these dependencies into a continuity risk.

Source: Info-Tech Research Group · Overlapping AI Outages Expose an Enterprise Resilience Gap

Read at the source →

Resilience and continuity ·

The hidden risk of relying on a single cloud platform

The article published by the BCI explains that cloud-native backup features are not enough when production and backups reside on the same platform, which creates a single point of failure. It proposes the 3-2-1 scheme (three copies, two media, one independent copy outside the platform) together with cybersecurity and continuity planning to sustain operations during provider disruptions.

Source: Business Continuity Institute (BCI) · Martin Lewis (Wavenet) · The Hidden Risk of Cloud Reliance

Read at the source →
News RSS feed

Shall we talk?

A managing partner will attend you personally. Free 30-minute executive assessment, in person or online.

Book your assessment +52 55 3626 2700