🚀 Medial Secures Investment on Shark Tank India - Fueling the Future of Professional Social Networking. 🔥

Startup Showcase

Premium Content

Try our Valuation Calculator →

News on Medial

Microsoft’s new safety system can catch hallucinations in its customers’ AI apps

The Verge

The Verge · 1y ago

Microsoft’s new safety system can catch hallucinations in its customers’ AI apps

Microsoft has introduced new safety features for its Azure AI customers. The features, available on its platform, include Prompt Shields, Groundedness Detection, and safety evaluations. The Prompt Shields tool blocks malicious prompts, while Groundedness Detection identifies and blocks hallucinations. Safety evaluations assess vulnerabilities in models. The company is also planning to add features for directing safe outputs and tracking prompts to identify problematic users. The safety features will not only be attached to popular models like GPT-4, but users will also be able to manually point the features to smaller, open-source models in Azure's model garden.

Related News

Why AI makers don't tell their chatbots to answer only what they know

Business Insider

Business Insider · 1y ago

Why AI makers don't tell their chatbots to answer only what they know

According to Jared Kaplan, co-founder of Anthropic, chatbots should not be overly worried about making mistakes in order to remain useful to users. Kaplan believes occasional errors or "hallucinations" are a trade-off for an otherwise valuable AI system. While the ultimate goal is to have AI platforms with zero hallucinations, striking a balance between accuracy and practicality remains a challenge for the sector. Kaplan's company, Anthropic, focuses on AI safety and research with an emphasis on ethical values and safety concerns.

Anthropic summons the spirit of Flash games for the AI age

Arstechnica

Arstechnica · 2m ago

Anthropic summons the spirit of Flash games for the AI age

Anthropic has launched a feature expanding its Artifacts document management system to create and share AI-generated apps, reminiscent of early 2000s Flash games. Users can build apps using plain English within Claude's chatbot interface, thanks to a new API capability. The apps run on modern web code and are shared with links, requiring no installation. Claude handles the coding, allowing users to focus on ideas, providing a simplified, interactive development environment.

Cityflo implements full-stack driver safety intelligence systems across metro fleets

Economic Times

Economic Times · 3m ago

Cityflo implements full-stack driver safety intelligence systems across metro fleets

Cityflo, an urban mobility firm, has introduced an AI-powered driver safety system across its fleet to enhance road safety and fleet discipline. The new system combines an advanced driver assistance system (ADAS) and a driver monitoring system (DMS) to reduce risks and improve driver accountability. This initiative, in partnership with safety startup Cautio, aims to address the root causes of road accidents and is expected to achieve full fleet integration soon, enhancing operational excellence.

Businesses testing ways to safeguard against AI hallucination

Economic Times

Economic Times · 1y ago

Businesses testing ways to safeguard against AI hallucination

Businesses across various sectors, including telecom, medicine, and ad-tech, are facing the challenge of "AI hallucinations," where artificial intelligence produces inaccurate, biased, or unsuitable content or choices. These hallucinations can result in loss of business, threats to life, increased costs, and diminished trust in AI's output. Telecom companies may experience security lapses and network outages, while ad-tech platforms may produce material that is not aligned with brand values. Safeguarding against these hallucinations requires prompt engineering techniques, grounding models with factual data, and incorporating human supervision in AI applications.

One month with Microsoft’s AI vision of the future: Copilot Pro

The Verge

The Verge · 1y ago

One month with Microsoft’s AI vision of the future: Copilot Pro

Microsoft's Copilot Pro subscription, priced at $20 per month, offers AI-powered features within Office apps, priority access to OpenAI models, and enhanced image generation. The improved image creation tool, Designer, utilizes the DALL-E 3 model to generate high-detail widescreen images. While Designer's results can be impressive, it is not without its flaws, occasionally inserting random elements or struggling with text. Copilot Pro also brings AI capabilities to Word, PowerPoint, Excel, and Outlook, allowing for text generation, visualizations, and email summarization. The subscription shows potential, particularly for image generation, but its value proposition may vary depending on individual needs.

Apple will revamp Siri to catch up to its chatbot competitors

Economic Times

Economic Times · 1y ago

Apple will revamp Siri to catch up to its chatbot competitors

Apple is reorganizing its AI strategy and focusing on generative artificial intelligence (AI) to improve its virtual assistant, Siri. The company wants to catch up in the AI race and has made generative AI a key project. The updated Siri will be more conversational, versatile, and capable of following a conversation. Apple also plans to apply generative AI throughout its business, increase memory in iPhones to support Siri's new capabilities, and potentially license AI models from other companies. The company's focus on AI is driven by the fear of losing dominance in the smartphone market and competition from AI-powered apps.

Microsoft’s Office apps now have free Copilot Chat features

The Verge

The Verge · 2d ago

Microsoft’s Office apps now have free Copilot Chat features

Microsoft has introduced a free Copilot Chat sidebar in its Office apps, including Word, Excel, PowerPoint, Outlook, and OneNote, for business Microsoft 365 users. This feature, integrated into the Microsoft 365 apps, aids in drafting documents, analyzing spreadsheets, and more. It offers content-aware AI capabilities at no additional cost. However, users with a $30 per month Microsoft 365 Copilot license still receive premium features and priority access to advanced integrations like GPT-5.

Users find that Facebook’s new AI stickers can generate Elmo with a knife

Arstechnica

Arstechnica · 1y ago

Users find that Facebook’s new AI stickers can generate Elmo with a knife

Users are misusing Meta's new AI-generated stickers feature on Facebook Messenger, creating and sharing awkward images of copyright-protected characters. Artists have posted viral series of virtual stickers, including depictions of Mickey Mouse with a machine gun, the flaming Twin Towers, and Elmo brandishing a knife. While Meta has implemented filters to catch potentially offensive content, some combinations are slipping through. This highlights the challenges faced by moderation teams as AI-powered apps become more prevalent. Meta spokesperson Andy Stone admitted that generative AI systems can produce inaccurate or inappropriate outputs and pledged to improve the features based on user feedback.

AI is confusing — here’s your cheat sheet - The Verge

The Verge

The Verge · 9m ago

AI is confusing — here’s your cheat sheet - The Verge

Artificial intelligence (AI) is a complex and rapidly evolving field. In order to understand the latest developments, it is important to be familiar with common AI terms. AI refers to computer systems that can think like humans, but it is often used as a buzzword. Machine learning involves training AI systems to make predictions based on data. Artificial general intelligence (AGI) refers to AI that is as smart as or smarter than humans. Generative AI can create new text, images, and code. Hallucinations occur when AI tools make up responses based on training data. Bias can arise in AI systems due to the biases in the training data. Other terms include AI models, natural language processing (NLP), inference, tokens, neural networks, transformers, and more. AI systems are trained on data and parameters are variables that determine how the system responds. Hardware such as Nvidia's H100 chip and neural processing units (NPUs) are used to run AI systems. Companies like OpenAI, Microsoft, Google, Meta, Apple, and Anthropic are developing AI tools and models.

Yango ride-hailing service introduced new safety features for UAE

Startup News FYI

Startup News FYI · 1y ago

Yango ride-hailing service introduced new safety features for UAE

Yango, the international ride-hailing service, has launched new safety features for its users in the UAE. These features include speeding control and driving-style monitoring technologies aimed at enhancing road safety and reducing the risk of accidents. Yango's commitment to advocating road safety aligns with the UAE's vision for road safety as the country continuously updates traffic laws and implements advanced technology to improve road safety. Yango is also working on an AI-driven route monitoring system to further enhance the safety of its rides.

Trackers

Active Indian VC’s

OG Capital Email

With a hands-on approach, OG Capital aims to invest in over 20 promising...

Accel Partners Email

Early and growth-stage investments in disruptive technology companies with...

Early-stage venture capital firm investing in technology startups in India. Focus on...

Access All Trackers

Startup Showcase Winners

Aug 2025

Build. Comply. Fund. Fly.

Crypto to INR payments app for Indians

Your CV, Done Right.

Enter Ongoing Startup Showcase

Top Users

Trending News on Medial

GST 'fraud': MNCs gained from fak ...

Exclusive: Lighthouse-backed Duro ...

US tariffs to shear 5-10% off hom ...

Download the medial app to read full posts, comements and news.