Tech Giants Agree To US Government AI Model Testing, FTC Bans Kochava Location Data Sales

Google Microsoft And xAI Agree To Let US Government Test AI Models

• Major firms including OpenAI and Anthropic join pre release testing pact.
• Companies renegotiated deals with Washington for safety evaluations.
• Agreement covers models before public rollout to address risks.

WASHINGTON D.C., May 05 (TNGB) – Google, Microsoft, and xAI reached agreements permitting U.S. government evaluation of their artificial intelligence models prior to public release. OpenAI and Anthropic participated after renegotiating prior arrangements with federal authorities. The framework establishes structured testing to identify safety and security issues in advanced systems before they become widely available.

The voluntary pacts reflect growing emphasis on pre deployment risk assessment.

Why This Matters: Government access to models before launch can identify safety concerns and shape responsible deployment standards across the industry.

Chrome Now Lets Users Share Approximate Location Instead Of Precise Gps

• Android version introduces option to limit data sent to websites.
• Feature reduces tracking while still enabling location based services.
• Update addresses long standing privacy complaints about stalking.

USA, May 05 (TNGB) – Google introduced a privacy control in Chrome for Android that allows users to share approximate rather than precise location data with websites. The setting gives individuals greater choice over the granularity of geolocation information transmitted during browsing. It responds to long standing concerns about excessive tracking through exact GPS coordinates requested by web pages.

Users can adjust the setting on a per site basis through browser controls.

Why This Matters: Granular location controls in widely used browsers reduce unnecessary data exposure while preserving functionality for location aware services.

Brussels Criticizes Meta For Weak Id Checks And Proposes Stronger Fixes

• EU officials say current age verification falls short of requirements.
• Options include behavioral profiling or mandatory government ID linkage.
• Proposal would turn every login into a surveillance checkpoint.

BRUSSELS, May 05 (TNGB) – European Commission officials determined that Meta’s current identity verification measures for age restricted content remain insufficient. Regulators are considering behavioral age estimation technology or direct linkage to government issued identification as potential remedies. Both approaches would convert routine logins into more intensive data collection or verification events.

The proposals form part of broader efforts to enforce online safety rules.

Why This Matters: Stricter identity requirements on major platforms could reduce anonymity and expand data collection tied to everyday digital access.

UK Google DeepMind Staff Seek To Block Ai Use In Military Applications

• Employees at AI research lab push for internal policy restrictions.
• Vote aims to prevent company models from supporting defense contracts.
• Move reflects growing ethical concerns among technologists.

LONDON, May 05 (TNGB) – Staff at Google’s DeepMind artificial intelligence research laboratory in the United Kingdom voted to advocate for internal policies that would restrict use of company models in military contexts. The effort seeks formal limits on participation in defense related projects. Employees cited ethical considerations surrounding autonomous systems and potential weapon applications.

The internal action highlights tensions between commercial AI development and military use cases.

Why This Matters: Employee advocacy within leading AI labs can influence corporate boundaries on sensitive applications of the technology.

Hackers Steal Student Data In Breach At Education Tech Firm Instructure

• Attackers accessed records from platform used by schools and universities.
• Incident affects Canvas and other widely adopted learning management tools.
• Company notified affected institutions about the data compromise.

USA, May 05 (TNGB) – Education technology provider Instructure experienced a data breach that exposed student records stored on its systems. The incident impacted the Canvas learning management platform and related services used by thousands of schools and universities. The company confirmed the theft and began notifying educational partners regarding the scope of compromised information.

Affected institutions are assessing notification obligations and potential risks to students.

Why This Matters: Breaches at education technology platforms can expose large volumes of student personal data and prompt reviews of vendor security practices.

Ftc Moves To Ban Data Broker Kochava From Selling Americans Location Data

• Federal agency targets sale of precise geolocation information.
• Action follows years of complaints about unchecked data practices.
• Ban would prohibit transfer of sensitive mobile device tracking details.

WASHINGTON D.C., May 05 (TNGB) – The Federal Trade Commission initiated action to prohibit data broker Kochava from selling precise location data collected from American consumers. Officials cited ongoing privacy violations and the elevated risks associated with granular mobile device tracking. The proposed ban would prevent further commercial transfer of such sensitive information.

The proceeding may establish stronger precedents for location data protection.

Why This Matters: Restrictions on data broker sales of precise location information can limit commercial surveillance and reduce risks of data misuse.

Openai Claims New Chatgpt Default Model Hallucinates Way Less

• Company switched default to GPT 5.5 Instant for improved accuracy.
• Update delivers more reliable answers with reduced fabrications.
• Feature also limits unnecessary emoji usage in responses.

USA, May 05 (TNGB) – OpenAI announced that GPT 5.5 Instant now serves as the default model for ChatGPT and produces significantly fewer hallucinations than prior versions. The company reported extensive internal testing before deployment to improve factual consistency and reduce unnecessary embellishments. Users may observe more precise and contextually appropriate replies across interactions.

The update forms part of ongoing efforts to enhance model reliability.

Why This Matters: Reductions in model hallucinations can increase user confidence in AI generated information for everyday and professional tasks.

Next Call Of Duty Will Skip Ps4 And Xbox One Consoles

• Activision determined modern features exceed last generation hardware capabilities.
• Development focuses exclusively on current generation platforms and PC.
• Long time players on older consoles face limited options for future titles.

USA, May 05 (TNGB) – Activision Blizzard confirmed that the next Call of Duty title will not be released on PlayStation 4 or Xbox One. The studio determined that advanced features and performance requirements surpass the technical limits of last generation consoles. Development resources concentrate on PlayStation 5, Xbox Series X and S, and personal computer versions.

The decision accelerates the industry’s transition away from legacy platform support.

Why This Matters: Ending support for previous generation consoles pushes players toward hardware upgrades and narrows access for those without current systems.

Huawei Expects 12 Billion Dollars In Ai Chip Revenue From Domestic Demand

• Chinese firm benefits as local AI model development accelerates.
• Nvidia market share in China reportedly drops near zero due to restrictions.
• Domestic fabs struggle to meet surging production needs.

CHINA, May 05 (TNGB) – Huawei projected approximately 12 billion dollars in revenue from AI chip sales this year, driven by strong domestic demand for homegrown artificial intelligence models. The company gains as U.S. export restrictions reduce Nvidia’s market share in China to near zero. Local semiconductor manufacturing facilities operate at capacity yet still cannot fully satisfy rising requirements.

The forecast illustrates accelerating self reliance in China’s technology sector.

Why This Matters: Growth in domestic AI chip production under export constraints can reshape global semiconductor supply chains and technology competition.

Congress Extends Controversial Surveillance Law For 45 Days

• Lawmakers approve temporary continuation of Section 702 powers.
• Extension buys time before full expiration and debate on reforms.
• Program allows warrantless collection of foreign intelligence data.

WASHINGTON D.C., May 05 (TNGB) – Congress approved a 45 day extension of the controversial Section 702 surveillance authority hours before its scheduled expiration. The temporary measure maintains existing capabilities for collecting foreign communications while lawmakers continue negotiations on longer term reforms and oversight adjustments. The program permits warrantless acquisition of data that may incidentally include information about U.S. persons.

The short term renewal avoids immediate disruption to intelligence operations.

Why This Matters: Temporary extensions of surveillance authorities sustain intelligence collection while Congress weighs competing security and privacy considerations.