salesforce announces anthropic

Key Takeaways

• Rochester Institute of Technology, Idaho State University, and the University of Florida have launched new AI degree programs to meet increasing demand.
• Idaho State University's new AI bachelor's degree prepares students for a field projected to grow by 34 percent by 2034.
• Salesforce is introducing Agentic Work Units (AWUs) to measure AI's business impact, moving beyond token consumption.
• Reid Hoffman suggests tracking employee AI token usage ('tokenmaxxing') as an engagement metric.
• Duolingo reversed its plan to evaluate employees based on AI use, prioritizing job outcomes.
• Anthropic's new AI model, Claude Mythos Preview, raises cybersecurity concerns due to its 'vibe coding' capabilities.
• The U.S. Treasury Department is seeking access to Anthropic's Mythos AI model to identify potential vulnerabilities.
• Crush Security launched an AI-powered cybersecurity intelligence platform, raising nearly $3 million from investors.
• Monjur, Inc. appointed Tracy Thibodeau as its new Vice President of Product & AI to lead product strategy and AI initiatives.
• The African Union is developing a Common African Position on AI and an African Charter to guide its responsible use in governance, peace, and security.

RIT launches AI bachelor's degree amid growing demand

Rochester Institute of Technology (RIT) will offer a new Bachelor of Science degree in artificial intelligence starting this fall. This program aims to teach students how to design, develop, and deploy AI systems. Employers are actively seeking individuals with AI experience, making this degree timely and valuable. RIT will also introduce a minor in AI. The new program reflects the increasing integration of AI into daily life and the growing interest in the field.

UF Engineering introduces Master of Science in AI Systems

The University of Florida (UF) College of Engineering now offers a Master of Science degree in artificial intelligence systems, joining a national trend of AI-focused higher education. Launched in Fall 2024, the two-year program requires 18 credits of core courses in machine learning, AI systems, sensing, analysis, security, ethics, and deep learning, plus nine elective credits and a capstone project. The curriculum is designed to be hands-on, preparing students for careers in the rapidly evolving AI field. The program aims to create interdisciplinary AI experts by allowing students to explore AI applications across various engineering disciplines and other fields.

Idaho State University adds AI bachelor's degree

Idaho State University (ISU) will launch a new bachelor's degree program in artificial intelligence sciences this fall to meet rising demand. This program is expected to prepare students for careers in a field projected to grow by 34 percent by 2034. The degree will be offered jointly by the Departments of Mathematics and Statistics and Computer Science, with students choosing between concentrations in theoretical foundations or modern applications. The program may eventually expand into interdisciplinary areas like healthcare and business, with up to 80 percent of coursework available online.

Reid Hoffman supports tracking AI token usage

Reid Hoffman believes companies should track employee AI token usage as a way to gauge engagement with AI tools. He suggests this metric, termed 'tokenmaxxing,' can be a useful dashboard indicator, though not a perfect measure of productivity. Hoffman advises pairing token usage data with an understanding of what employees are using AI for, acknowledging that experimentation, even if it fails, is valuable. He also recommends embedding AI across organizations and holding regular check-ins to share learnings and successes.

Salesforce introduces new AI metric to measure results

Salesforce is introducing a new metric called Agentic Work Units (AWUs) to measure AI productivity, moving away from simply tracking token consumption. AWUs focus on the actual output and business impact of AI, rather than the amount of data processed. For example, Singapore Airlines uses AWUs to track customer service issue resolution times, and Williams Sonoma uses it for product recommendation effectiveness. This approach counters the trend of 'tokenmaxxing,' which Salesforce argues is a vanity metric, emphasizing that the value of AI lies in the work it accomplishes.

Crush Security launches AI platform for cybersecurity tool selection

Crush Security has launched its AI-powered cybersecurity intelligence platform, aiming to transform how security tools are evaluated and purchased. The startup has raised nearly $3 million from investors, including CISOs, to help customers make better purchasing decisions. Crush Security positions itself as a solution provider using AI and data normalization to compare tools, identify gaps, and automate the reseller process. Its AI advisor, Cassandra, can answer natural-language questions to help customers understand their security needs and compare vendors.

Monjur appoints Tracy Thibodeau as VP of Product & AI

Monjur, Inc. has appointed Tracy Thibodeau as its new Vice President of Product & AI. In her new role, Thibodeau will lead the company's product strategy and artificial intelligence initiatives. Her focus will be on accelerating innovation within Monjur's legal technology platform, including its AI-powered assistant, Monjur Pilot. Monjur provides attorney-supervised contract intelligence solutions for managed service providers (MSPs).

Duolingo reverses AI performance review plan after employee pushback

Duolingo's CEO, Luis von Ahn, announced the company has reversed its plan to evaluate employees based on their use of AI in performance reviews. Employees expressed concern that the company was pushing AI for its own sake rather than focusing on job outcomes. Von Ahn stated that the primary focus will remain on employees performing their jobs as well as possible, with AI being a helpful tool but not a mandatory requirement. This decision follows a previous backtrack on an 'AI-first' company strategy.

New AI model Mythos poses risks to internet security

Anthropic's new AI model, Claude Mythos Preview, raises concerns about cybersecurity threats, with similar capabilities expected from other labs within 6 to 18 months. The ease with which AI can now generate code, known as 'vibe coding,' allows untrained individuals to create software applications. However, these applications may lack security reviews, leaving them vulnerable to data breaches or system takeovers. The increasing accessibility of AI for both coding and finding flaws threatens the security of various internet services.

African Union discusses AI's role in governance and security

The African Union (AU) Peace and Security Council convened a ministerial session on April 16, 2026, to discuss Artificial Intelligence (AI) concerning governance, peace, and security in Africa. The meeting addressed the rapid development and deployment of AI across various sectors, highlighting the need to harness its benefits while mitigating risks. Challenges such as infrastructure, digital literacy, unreliable electricity, and fragmented data regulations were discussed. The AU aims to develop a Common African Position on AI and an African Charter to guide its responsible use.

GSA aims to automate one million work hours with AI

The General Services Administration (GSA) is working to automate one million work hours using its internal AI tool, USAi, following a significant reduction in its workforce. The agency has already automated about 400,000 hours, focusing on non-high-value tasks to allow employees to focus on more critical work. This initiative, part of an 'eliminate, optimize and automate' strategy, began internally but may expand beyond GSA if successful. The project involves a dedicated cohort of GSA employees identifying and solving automation challenges.

Autodesk proposes AI policies for design and manufacturing

Autodesk has released public policy recommendations to support the responsible adoption of artificial intelligence (AI) in the design and make industries, including architecture, engineering, construction, and manufacturing. The recommendations focus on accelerating digital tool adoption, investing in workforce development, fostering data sharing, and supporting responsible AI deployment. Autodesk believes these policies will help these industries drive economic growth by leveraging industrial AI to improve decision-making, reduce waste, and enhance resilience. The company emphasizes the need for partnership between the public and private sectors to maximize AI's benefits.

US Treasury seeks access to Anthropic's Mythos AI model

The U.S. Treasury Department's technology team is seeking access to Anthropic's Mythos AI model to identify potential vulnerabilities. Chief Information Officer Sam Corcos directed the cybersecurity team to prepare for threats from advanced AI systems. Anthropic is releasing the Mythos model to a limited number of institutions. This move by the Treasury highlights concerns about the security implications of powerful new AI models and the need for proactive vulnerability assessment.

Sources

• As AI use booms, RIT offers bachelor's degree in artificial intelligence
• UF Engineering offers a master’s in AI. Here’s what to know.
• Idaho State University to offer artificial intelligence degree
• Reid Hoffman weighs in on the 'tokenmaxxing' debate
• Exclusive: Salesforce takes on 'tokenmaxxing'
• Crush Security Exits Stealth Seeking To Become AI-Powered Trusted Advisor Of The Future
• Monjur Appoints Tracy Thibodeau as Vice President of Product & AI
• Duolingo was evaluating its workers' AI use. Workers pushed back.
• Opinion | After Mythos, the Future of the Internet Is At a Crossroads
• Artificial Intelligence: Governance, Peace and Security in Africa
• GSA looks to automate a million work hours, after losing nearly 40% of its workforce
• Autodesk shares AI Public Policy Recommendations to help accelerate innovation, efficiency, and resilience in Design and Make industries
• US Treasury Seeking Access to Anthropic’s Mythos to Find Flaws