KIMP’s Picks June 12th: Weekly Highlights In Design, Marketing, AI, & More

Are you someone who scrolls through social media feeds to catch up on the latest buzz? Innovative campaigns from brands, major AI news causing ripples, social media algorithm changes you probably missed – all the highlights wrapped up in a single place. Welcome to KIMP’s Picks’ June 12th edition.

Ready to get a sneak peek into this week’s buzz? Let’s get started.

In the Marketing Realm 

Brand trust matters now more than ever 

ADWEEK shared a post highlighting a new global study from McCann. The study highlights a deepening trust problem between consumers and brands. Based on responses from over 20,000 people across 20 markets, the findings show that trust is becoming a key factor in whether people stay loyal or leave.

The research reveals that 69% of consumers say they have abandoned brands they no longer trust, while 80% say they actively choose brands they do trust. At the same time, 55% believe brands were more truthful two decades ago, pointing to a long-term decline in perceived honesty.

AI is adding pressure to this issue. Around 76% of respondents worry they will soon struggle to tell real human content from AI-generated material. More than half also say transparency around AI use in advertising is essential for building trust.

The message here is clear that trust is now a deciding factor in retention, and transparency, especially around AI, is becoming non-negotiable for long-term loyalty.

Burger King’s unique campaign for FIFA 

Burger King France is leaning into a simple truth about football nights. Most fans default to pizza while watching matches. Instead of fighting that habit, the brand is choosing to adapt to it with a new campaign built around acceptance rather than competition.

The campaign introduces limited-edition square boxes filled with Burger King’s “Baby Burgers,” miniature versions of its core products like the Whopper. The idea is positioned as a playful twist on the traditional pizza delivery format, aligning directly with how people already behave during major football moments. 

The messaging is straightforward. If audiences are already ordering pizza during matches, Burger King wants to be part of that moment instead of trying to change it. The brand builds on the fact that attention is less about changing consumer habits and more about fitting into them in a way that feels natural and timely.

In the Content Marketing Realm 

How people use AI to search for answers 

New survey data from Stella Rising challenges common assumptions about how users interact with AI search tools. Across two studies in 2025 and 2026, most people were found to still use short, keyword-style prompts rather than long, highly engineered instructions often seen in marketing discussions around AI search.

In the January 2026 survey, two-thirds of users wrote prompts under 15 words. Many were just simple searches like “best hiking shoes” or “Nike tennis shoes.” Only a small minority used complex, structured prompting styles. This closely mirrors traditional search behavior and suggests AI tools are being used more like Google than many expect.

The data also shows common patterns like price-based queries, “best” searches, and location-driven intent, reinforcing that commercial behavior is strong inside AI tools.

What stands out most is how personal context changes outcomes. Users are also sharing details they would never include in a search engine, which directly influences which brands are surfaced in recommendations. 

For brands, this shift means visibility is no longer just about keywords. It now depends on how well content matches real human situations, not just search terms.

Most AI citations do not seem to bring brand visibility 

A new study by Semrush and Kevin Indig reveals a major gap in AI search visibility. The study found that being cited by AI does not necessarily mean your brand gets recognized. After analyzing nearly 4,000 domain appearances across ChatGPT, Gemini, Google AI Overviews, and Google AI Mode, researchers found that 62% of citations were “ghost citations” where content was used as a source, but the brand name was never mentioned in the response.

The study also found significant differences between AI platforms. ChatGPT heavily relies on citations but rarely mentions brands, while Gemini is far more likely to name brands directly. Strong consumer brands such as Google and Apple were mentioned more often than cited, whereas publishers and research sites were typically cited without receiving brand recognition.

This highlights a key detail – success in AI search will depend more on earning brand mentions alongside citations, as mentions drive awareness while citations alone often leave brands invisible to users.

In the Design Realm 

Canva’s Magic Layers brings editable design directly into AI assistants 

Canva has launched its Magic Layers feature inside Google Gemini and OpenAI’s ChatGPT. This bridges a long-standing gap between AI image generation and real-world design work. The feature turns static AI-generated images into fully editable Canva files with layers, text, and objects that can be modified directly.

The core problem it targets is that AI tools can quickly generate high-quality visuals, but those outputs usually arrive as flat images that are difficult to edit or repurpose. Magic Layers changes that by breaking images into structured components, making it possible to adjust layouts, edit text, or reposition elements without starting from scratch.

The feature is now embedded directly into ChatGPT and Gemini workflows. So users can generate an image in either platform and instantly convert it into an editable Canva design, removing steps like downloading and re-uploading files.

Visual identity made tangible 

The Brand Identity shared a post exploring the brand identity created for Belle & George. At the core of the identity is a hand-drawn wordmark shaped by intuition. 

The imperfections and gestures in the lettering are treated as part of the meaning, not something to be corrected.

A central feature of the identity is the ambigram structure, designed so the mark reads across orientations. This idea came from experimentation with symmetry and reflection, inspired by artistic techniques that test composition through inversion. The result is a system that feels deliberately crafted, where balance is achieved through manual iteration rather than automated symmetry tools.

In an AI-heavy creative landscape, hand-driven design choices are becoming a form of differentiation. Imperfection, authorship, and visible craft are effective signals of trust and identity.

In the AI Realm 

Anthropic launches Claude Fable 5 with next-gen capabilities

Claude has introduced two new models, Fable 5 and Mythos 5. These two models signal a major jump in capability across software engineering, research, vision, and long-context reasoning. The company describes Fable 5 as its most advanced general-use model yet, with strong performance across complex tasks, especially those requiring long, multi-step thinking.

Alongside the performance leap comes a strong safety framework. Certain sensitive queries are automatically redirected to an earlier model, Claude Opus 4.8, rather than being handled directly by Fable 5. These safeguards are designed to reduce misuse risk in areas like cybersecurity and biology, even if that means occasional over-blocking of harmless requests.

Google’s DiffusionGemma rethinks text generation 

Google has released DiffusionGemma, an experimental open model designed to rethink how large language models generate text. Instead of producing words one by one like traditional AI systems, it generates entire blocks of text in parallel, offering up to four times faster output on supported GPUs.

The model is built on a 26B Mixture of Experts architecture, but only a small portion of its parameters are active at any time. Its key innovation is a diffusion-based approach, similar to how image generators refine visuals from noise, but applied to text. The system starts with a full canvas of placeholder tokens and iteratively refines them until a coherent output emerges.

This parallel generation method changes the core interaction pattern. Rather than waiting for sequential token-by-token responses, the model processes up to 256 tokens at once. 

Google introduces Gemini 3.5-powered Live Translate 

Google has introduced Gemini 3.5 Live Translate, a new audio model designed for real-time speech-to-speech translation across more than 70 languages. The system focuses on making conversations feel natural by preserving tone, pacing, and intonation while translating speech on the fly.

Unlike traditional translation tools that wait for a speaker to finish before responding, this model works continuously. It listens and translates in near real time, staying only a few seconds behind the original speaker. This reduces pauses and makes conversations flow more naturally across languages.

The technology is already being rolled out across Google’s ecosystem, including the Gemini Live API for developers, Google AI Studio, Google Meet for enterprise users, and the Google Translate app on Android and iOS. It also supports multilingual communication across thousands of language pair combinations.

View on Threads

Google upgrades NotebookLM with new research, coding and other capabilities 

Google has rolled out major upgrades to NotebookLM. The new features position NotebookLM as a more advanced AI research and analysis assistant powered by Gemini 3.5 and Antigravity. The update focuses on deeper reasoning, better reliability, and the ability to handle complex, multi-step research tasks.

The biggest change is the shift toward agent-like behavior. Each notebook now includes a secure cloud environment where the system can write and run code, helping users analyse data, explore documents, and generate structured insights instead of just summaries.

NotebookLM also expands into multi-format creation. Users can now turn research into outputs like PDFs, spreadsheets, presentations, charts, and structured data files. Another key upgrade is how users start research. Instead of requiring pre-loaded sources, NotebookLM can now actively help gather material, suggest references, and even pull relevant information from the web through integrated search. This lowers the barrier for open-ended research tasks.

Claude introduces managed agents with scheduling and secure vault-based tool access 

Anthropic has upgraded its Claude Platform with new Managed Agents that can now run on schedules and securely access external tools through environment variables stored in vaults. The update is currently in public beta and is designed to make AI agents more autonomous in handling recurring and system-level tasks.

A key addition is scheduled execution. Developers can now assign a cron-style schedule to an agent, allowing it to automatically run at set intervals without needing custom infrastructure. Each run starts a fresh session where the agent completes its assigned task independently. This is positioned for repetitive workflows like daily reports, compliance checks, or data monitoring.

Social Media Feature Updates and Algorithm Changes 

Snapchat tightens teen sharing rules with friends-only profiles for ages 13-15

Snapchat is rolling out a new set of protections that reshape how younger teens share content on the platform. Users aged 13 to 15 will now have dedicated profiles where they can create, save, and showcase Stories and Spotlight videos, but only to mutually accepted friends.

The key change is the removal of public distribution for this age group. Content from younger teens will no longer appear to non-friend audiences on Spotlight, marking a clear shift toward private, closed-circle sharing instead of broader visibility.

These profiles are designed to focus on creativity within trusted networks. There are no public metrics like view or favourite counts, and content is visible only to confirmed friends, not followers or wider audiences. 

Snapchat says the update is part of a tiered system based on age. Additionally, teens aged 16 to 17 will have limited, optional access to public sharing with added safeguards, while users 18 and above retain full public posting capabilities.

Instagram finally lets you manually rearrange photos in your grids

Instagram is rolling out a long-requested update that allows users to manually rearrange posts on their profile grid. The feature gives people direct control over how their profile appears, moving beyond the traditional chronological or fixed layout system.

Users can now tap any post on their profile, select “Reorder grid,” and enter an editing mode where photos and videos can be dragged into a new arrangement. Once satisfied, they simply exit the screen and the updated order is saved to their profile.

The feature has been in testing for some time and is now starting to roll out more widely, though availability may vary as it reaches users in stages.

Instagram will now let users “talk to the algorithm”

Instagram is rolling out a new feature called “Your Algorithm”, giving users direct visibility into the topics the platform believes they are interested in, along with the ability to adjust those preferences. The update is now expanding across Feed, Reels, and Explore.

The feature shows a list of inferred interest areas based on user behaviour. People can add topics they want to see more of or remove ones they want less of, effectively giving them more control over how their feed is shaped.

The system is powered by newer AI models that can translate complex content patterns into understandable topics. This makes it easier for users to interpret what the algorithm is doing and adjust it in real time. 

View on Threads

LinkedIn announces Creator Marketplace and BrandWorks 

LinkedIn is introducing two new products, Creator Marketplace and BrandWorks, aimed at helping brands build more credible and effective B2B marketing campaigns. The update is designed around a central idea that in B2B, trust and the voice delivering a message now matter as much as the message itself.

Creator Marketplace brings creators and brands into a single system where they can discover, evaluate, and collaborate more easily. Brands can search for vetted creators based on expertise, review audience and performance signals, and turn both organic and paid creator content into amplified campaigns. 

BrandWorks focuses on campaign execution. It provides hands-on support for strategy, creative development, and content production tailored specifically for LinkedIn’s professional audience. The goal is to help brands turn insights into campaigns that better match how decision-makers consume content on the platform.

Let’s Wrap Up…

That’s it for this week’s roundup. Stay tuned as we keep tracking the biggest shifts across AI, design, marketing, and social platforms. Each edition brings you the updates that actually matter, all in one place. We’ll be back next week with more highlights, sharper insights, and the stories shaping what comes next.

See you in the KIMP’s Picks edition.